F. Mondada, E. Franzi, and P. Ienne, "Mobile robot miniaturisation: A tool for investigation in control algorithms," in Proc. of the 3rd Int. Symp. on Experimental Robotics. Heidelberg: Springer-Verlag, 1994, pp. 501--513.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....5. Finally, section 6 concludes and points at future work. 2. Previous Research Recently we have studied several different aspects of the evolution of controllers for boxpushing. In our studies we employed a publicly available mobile robot simulator based on the widely used mobile robot Khepera [9]. A schematic illustration of the robot is shown in figure 1. The (simulated) Khepera is equipped with eight proximity sensors and two motors. We augmented the simulator with a movable object, i.e. a circular box. The box pushing task was introduced by Lee, Hallam and Lund [7] Our work differs ....

F. Mondada, E. Franzi and P. Jenne. Mobile robot miniaturisation: A tool for investigating in control algorithms. In T. Yoshikawa and F. Miyazaki, eds., Proceedings of the Third International Symposium on Experimental Robotics, pp. 501-513. Springer-Verlag, Berlin, 1993.

....in section 3 and discussed in section 4. Finally, section 5 concludes and points at future work. 2 Experimental Procedure 2. 1 The robot simulator and previous research In our box pushing studies we employ a publicly available mobile robot simulator based on the widely used mobile robot Khepera [5]. A schematic illustration of the robot is shown in figure 1. The (simulated) Khepera is equipped with eight proximity sensors and two motors. We augmented the simulator with a movable object, i.e. a circular box. One of our studies focused on the type of fitness function to be used for ....

F. Mondada, E. Franzi and P. Jenne. Mobile robot miniaturisation: A tool for investigating in control algorithms. In T. Yoshikawa and F. Miyazaki, editors, Proceedings of the Third International Symposium on Experimental Robotics, pp. 501-513. Springer-Verlag, Berlin, 1993.

....the neural control mechanism, and the evolutionary algorithm (EA) As part of the EA the di erent tness functions considered are described. 2. 1 Simulation environment In our experiments, we employed a publicly available mobile robot simulator 1 based on the widely used mobile robot Khepera [10]. We augmented the simulator with a movable object, i.e, the circular box shown in gure 1. The main adaptations in the source code concerning the movable object can be found in appendix B. The left panel in gure 1 illustrates the experimental set up for studying box pushing behaviour. The ....

F. Mondada, E. Franzi, and P. Jenne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In T. Yoshikawa and F. Miyazaki, editors, Proceedings of the Third International Symposium on Experimental Robotics, pages 501-513, Berlin, 1993. Springer-Verlag.

....in section 3 and discussed in section 4. Finally, section 5 concludes and points at future work. 2 Experimental Procedure 2. 1 The robot simulator and previous research In our box pushing studies we employ a publicly available mobile robot simulator based on the widely used mobile robot Khepera [4]. A schematic illustration of the robot is shown in figure 1. The (simulated) Khepera is equipped with eight proximity sensors and two motors. We augmented the simulator with a movable object, i.e. a circular box. One of our studies focused on the type of fitness function to be used for ....

F. Mondada, E. Franzi and P. Jenne. Mobile robot miniaturisation: A tool for investigating in control algorithms. In T. Yoshikawa and F. Miyazaki, editors, Proceedings of the Third International Symposium on Experimental Robotics, pp. 501-513. Springer-Verlag, Berlin, 1993.

....left (2) before moving backwards (3) to give it a better approach (4) to the target. 6 Related Work Constructing robots that are small, easily deployable, and yet can do useful work and operate reliably over long period of times has proven to be quite di cult. Most miniature robots have wheels [2, 10], others can jump [5] y [14] or swim [3] So far, their use has been limited to research laboratories. Our Scout robots promise to be among the rst miniature robots ready for eld deployment. To control a large group of robots, a software architecture must allow for distributed operations and ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III, Proc. of the 3rd Int'l Symposium on Experimental Robotics, pages 501-513, Kyoto, Japan, Oct. 1993. Springer Verlag, London.

....large snake [7] These are envisioned as being restricted to certain special environments, such as water or gas pipes in need of inspection. Miniature mobile robots (on the order of cubic centimeters in size) have been investigated but are limited to laboratory investigation of robotic algorithms [11]. Our scouts promise to be among the first miniature robots to be ready for field exploration. Even smaller robots (microrobots and nanorobots) are the subject of speculation but realized systems are still to appear [3] To provide small mobile robots with sufficient processing power, ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III, Proc. of the 3rd Int'l Symposium on Experimental Robotics, pages 501--513, Kyoto, Japan, Oct. 1993. Springer Verlag, London.

....used in the simulations. A population of neural networks (Rumelhart and McClelland, 1986) are evolutionarily trained to control a mobile robot designed to keep an arena clear by picking up trash objects and releasing them outside the arena. The organism is a miniature mobile robot (Khepera; Mondada et al. 1993; see Figure 1) which is supported by two wheels that allow it to move in various directions by regulating the speed of each wheel. In addition, the robot is provided with a gripper module with two degrees of freedom. The robot is also provided with eight infrared proximity sensors and an optical ....

Mondada, F., Franzi, E., and Ienne, P. 1993. Mobile robot miniaturisation: a tool for investigation in control algorithms. In Yoshikawa, T. and Miyazaki, F. (Eds.). Proceedings of the Third International Symposium on Experimental Robotics.

....describes the experimental procedure by discussing the simulation environment, the neural control mechanism, and the evolutionary algorithm. Simulation environment In our experiments, we employed a publicly available mobile robot simulator 1 based on the widely used mobile robot Khepera [10]. We augmented the simulator with a movable object (i.e. the circular box; 1 http: diwww.epfl.ch lami team michel khep sim index.html Figure 1: Screen shot of the simulator, with the newly implemented movable object (large black circle) and the environment in which the experiments were ....

F. Mondada, E. Franzi, and P. Jenne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In T. Yoshikawa and F. Miyazaki, editors, Proceedings of the Third International Symposium on Experimental Robotics, pages 501-513, Berlin, 1993. Springer-Verlag.

....goal of VLab is to support teaching of behavioral modeling by studying colored robotic paths in 2D space. These paths are generated by one or more simulated robots or vehicles , which can draw paint on the canvas while moving around. A simplified robot model, inspired from the Khepera robot (Mondada et al. 1993), was adopted for supporting the vehicle models developed and examined. Each vehicle s sensors perceive stimuli sources and are directly connected to its motors that control motion, without elaborate processing. The topology of the connections between sensors and motors as well as the properties ....

Mondada, F., E. Franzi, and P. Ienne (1993) Mobile robot miniaturisation: A tool for investigation in control algorithms. Proceedings Third International Symposium on Experimental Robotics, Kyoto, Japan, October.

....evolutionary adaptation techniques, hereafter referred to as adaptive neuro robotics, has become a standard methodology in bottom up AI research. 40 Figure 8: The Khepera, a wheeled miniature mobile robot commonly used in adaptive robotics research (manufactured by K Team SA; for details see Mondada et al. 1993). The model shown here is equipped with infrared sensors and a simple camera. The rest of this section is structured as follows: Subsection 4.4.1 discusses the role of adaptive neurorobotics as a form of radical connectionism from an AI and cognitive science perspective. Different adaptation ....

Mondada, Francesco, Franzi, E., and Ienne, P. (1993). Mobile robot miniaturisation: A tool for investigating in control algorithms. In Third International Symposium on Experimental Robotics, Kyoto, Japan.

....from both parents. The mutation process allows new characteristics to develop and can be compared to the copying errors in the example taken from Braitenberg. Nolfi (Nolfi and Parisi, 1995) presents evolutionary robot experiments where a neural control structure is evolved for a Khepera (Mondada et al. 1993) robot. The Khepera is a miniature robot that operates in an arena on a desk top, see Picture 2. In the experiments described by Nolfi the robot is equipped with distance sensors and a gripper. The task was to clean the environment by throwing small objects outside the arena. This task implies ....

Mondada, F., Franzi, E., and Ienne, P. (1993). Mobile robot miniaturisation: A tool for investigating in control algorithms. In Third International Symposium on Experimental Robotics, Kyoto, Japan.

....big (top) and small (bottom left) arenas respectively represented as patches of 0.1m and 0.07m diameter lying on the floor. Bottom right Division of the small arena into 80 zones. The experiments with local communication (section 4) are carried out in Webots [13] a 3 D simulator of the Khepera [15] robots. The experiments are realized in two circular arenas of 1 meter and 2 meters diameter respectively as shown in figure 1. The simulator gives a faithful representation of the Khepera robots [15] by introducing noise in the robots movements and sensors measurements as measured on the real ....

....local communication (section 4) are carried out in Webots [13] a 3 D simulator of the Khepera [15] robots. The experiments are realized in two circular arenas of 1 meter and 2 meters diameter respectively as shown in figure 1. The simulator gives a faithful representation of the Khepera robots [15], by introducing noise in the robots movements and sensors measurements as measured on the real Khepera robots. Each robot is provided with 9 infra red (IR) sensors (8 are used to detect other robots and the arena walls, the 9th IR is activated only by the walls and allows to distinguish between ....

F. Mondada, E. Franzi & P. Ienne, (1993), `Mobile Robot Miniaturisation: a Tool for Investigation in Control Algorithms', Proceedings of ISER'93, Kyoto, Japan, October 1993. Also at http://diwww.epfl.ch/lami/robots/K-family/.

....of the problem by modelling the system as a set of probabilistic equations. The model is used to determine explicitly correlations between the system s variables. The prediction of the probabilistic model are then compared to simulations realised in Webots [10] a 3 D simulator of the Khepera [12] robots. In this respect, our work follows current line of research which develops probabilistic models of multi agents systems, e.g. for representing biological systems (ants society [4] or engineering systems ( 17] The rest of this paper is divided as follows. Section 2 describes the ....

....results of the simulations and compares those to the predictions of the probabilistic model. Section 5 conclude the paper with a short summary of the results of our studies. 2 The simulations 2. 1 The experimental set up Simulations were carried out in Webots [10] a 3 D simulator of the Khepera [12] robots. Simulations use two circular arenas of 1 meter and 2 meters diameter respectively as shown in figure 1. The Khepera robot is round with a diameter of 5.5cm. Thus, we study the exploration strategy in environments which are 1600 and 400 times the robot s size. The simulator gives a ....

F. Mondada, E. Franzi & P. Ienne, (1993), `Mobile Robot Miniaturisation: a Tool for Investigation in Control Algorithms', Proceedings of ISER'93, Kyoto, Japan, October 1993. Also at http://diwww.epfl.ch/lami/robots/K-family/.

....arguments for using a topological map and then present a well known diffusion mechanism for processing both the input and output data of the PTM. The basic qualitative aspects of functioning of the PTM will be described. Then, we will propose results of an experiment implemented on a Khepera robot [MONDADA93], that illustrates the performances of the PTM in function of the importance given to topology information. In this section, we will review both the evidence for topological organization in biological nervous systems, their usefulness for building internal representations, and artificial models. ....

....realization of that. The eight sensors provide the input to the PTM, and a diffusion mechanism is applied to them. Three basic actions are used: turn 35 left, turn 35 right or move forward by about 1.5cm. Such small movements are required because the range of the infrared sensors is about 3 cms [MONDADA93]. The global network is depicted on fig. 5. Figure 5: The control neural network for learning obstacle avoidance. Input is coded as an array of eight vectors, each corresponding to an infrared sensor. For each sensor, the value is quantized and diffused. Sensorial situations are coded on a ....

Mondada F., Franzi, Ed., Ienne P., (1993) Mobile Robot Miniaturisation: A Tool For Investigation In Control Algorithms, Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Oct. 28-30.

....disappear ( are born and die ) continuously. The study evaluates the influence of parameters, such as the range of communication and the rate of death and birth of robots and objects on the robots learning performance. Further, the simulation will be implemented in a real set up of ten Khepera [39] robots. Our work showed the importance of behavioural capacities alongside cognitive ones for addressing the symbol grounding problem. Behavioural mechanisms which act as external attentional processes are required alongside general cognitive abilities of associativity. We proposed a ....

F. Mondada, E. Franzi & P. Ienne (1993), `Mobile Robot Miniaturisation: a Tool for Investigation in Control Algorithms', Proceedings of ISER'93, Kyoto, Japan.

....briefly in the following subsections, together with examples of their use in the selforganization of sign processes in artificial organisms. 30 Figure 4: The Khepera, a wheeled miniature mobile robot commonly used in adaptive robotics research (manufactured by K Team SA; for details see Mondada et al. 1993). The model shown here is equipped with infrared sensors and a simple camera. 4.1.1 Artificial Neural Networks For the understanding of the argument here it suffices to know that an ANN is a network of a (possibly large) number of simple computational units, typically organized in layers (cf. ....

Mondada, Francesco, Franzi, E., and Ienne, P. (1993). Mobile robot miniaturisation: A tool for investigating in control algorithms. In Third International Symposium on Experimental Robotics, Kyoto, Japan.

....setups and the obtained results, and presents some analysis of how networks of different architectures solve the tasks. The final section then summarises the paper and discusses the results. 2 Experiments 2. 1 Robot and Simulator The robot used in Nolfi s experiments was a Khepera robot [9] equipped with eight infrared proximity sensors and a gripper with two degrees of freedom (i.e. the arm can be moved up and down, and the hands can be opened and closed) The gripper also has a light barrier object sensor, which detects whether or not there is an object between the hands. For ....

F. Mondada, E. Franzi, and P. Ienne, Mobile robot miniaturisation: A tool for investigation in control algorithms, Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Japan, pp. 501 - 513, 1993.

....found. 2.2. 2 Khepera Xmorph Experiments with the microrobot Khepera (K team, Lausanne, Switzerland) were performed using the distributed simulation environment Xmorph [Verschure, 1997b] Figure 3: About here Khepera (Figure 3A) is a circular robot with a diameter of 55 mm and a height of 30 mm [Mondada et al. 1993]. The basic configuration consists of two modules; the base plate and the processor module. All modules are connected by an extension bus to allow easy expansion. The base plate constitutes the elementary interface to the real world; effectors and obstacle light detection. The robot uses two ....

Mondada, F., Franzi, E., and Ienne, P. (1993). Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III: Proceedings of the 3rd International Symposium on Experimental Robotics, Kyoto, Japan, October 28-30, 1993, pages 501--513. Berlin: Springer Verlag.

....big (top) and small (bottom left) arenas respectively represented as patches of 0.1m and 0.07m diameter lying on the floor. Bottom right Division of the small arena into 80 zones. The experiments with local communication (section 4) are carried out in Webots [16] a 3 D simulator of the Khepera [18] robots. In the first set of experiments (section 4) simulation studies are realized in two circular arenas of 1 meter and 2 meters diameter respectively as shown in figure 1. The Khepera robot is round with a diameter of 5.5cm. Thus, we study the exploration strategy in environments which are ....

....with global communication (section 5) are carried out in an square arena of 78 by 78 cm. A photo of the physical set up and a picture of its implementation in Webots is given in figure 2. The experiments are realized with 1 to 4 Khepera robots. A technical description of the robots can be found in [18]. In the experiments, the robots are provided with eight infra red sensors (six in the front and two in the back with which they perform obstacle avoidance, see section 2.3) They communicate via radio using a Motorola radio transceiver (418 MHz, 4800 Baud rate) In these experiments, the database ....

F. Mondada, E. Franzi & P. Ienne, (1993), `Mobile Robot Miniaturisation: a Tool for Investigation in Control Algorithms', Proceedings of ISER'93, Kyoto, Japan, October 1993. Also at http://diwww.epfl.ch/lami/robots/K-family/.

....(3) by an additional factor e Forget , which describes the relation between learning and unlearning. Dc ij Forget = e Forget l k f (c ij ) s j q kj ) s i q ki ) V(t) 5) REAL WORLD PROGRAMMING INTERFACE Currently, the robot that can be controlled with NNetView is Khepera TM [14]. The robot sensors include a CCD color camera, distance sensors and ambient light sensors. Moreover, the robot consists of a 2 DOF gripper which delivers sensory information concerning arm positions and information whether or not there is an object present in the gripper (Fig. 2) The camera ....

Mondada, F., Franzi, E., Ienne, P. (1993): Mobile robot miniaturisation: A tool for investigation in control algorithms. In: Proceedings of the Third International Symposium on Experimental Robotics. Kyoto, Japan.

....body and a larger wheel base that means more time steps to turn the robot can evolve. In other words, the upper limit of the size and base of a robot is constrained by the sensor range. IV. Evolving Auditory Sensor Morphology. We have developed a new piece of hardware for the Khepera roboti [19], namely ears (see Fig. 4) This hardware is reconfigurable, and will allow us to study the co evolution of controller and ears morphology. As an example consider the cricket. The male cricket Fig. 4. The Khepera robot with ears. The ears have programmable amplifiers, synthesizers, and mixers. ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III. Lecture Notes in Control and Information Sciences 200, pages 501--513, Heidelberg, 1994. Springer-Verlag.

....the use of forbidden sequences of actions to force the learning of the expected behavior. In section 6, several experiments allow to demonstrate the interest of forbidden sequences of actions. Concluding remarks are given in section 7. 2. The miniature robot KHEPERA Khepera is a miniature robot [3] having a diameter of 6 cm and a weight of 86 g (Fig. 1) Two independent wheels allow the robot to move around. The number of possible actions is reduced to four hundreds. Eight infra red sensors help the robot to perceive its environment. The detection range is between 5 and 2 cm. Sensor data ....

F. Mondada, E. Franzi and P. Ienne, "Mobile robot miniaturisation: A tool for investigation in control algorithms," Third International Symposium on Experimental Robotics , Kyoto, Japan, October 1993.

....network. Our first series of experiments did not necessitate such an operation, which demonstrates the reliability of the simulator concerning the behaviors we evolved. 3.1 Description of the robot 5 6 8 4 3 7 2 1 Fig. 3. Khepera (5 cm diameter) and its simulated counterpart Dedicated to Khepera [11], the simulated mobile robot includes 8 infrared sensors (small rectangles) allowing it to detect the proximity of objects in front of it, behind it, and to the right and the left sides of it. Each sensors return a value ranging between 0 and 1023. 0 means that no object is perceived while 1023 ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Third International Symposium on Experimental Robotics, Kyoto, Japan, October 1993. This article was processed using the L A T E X macro package with LLNCS style

....the window position 3 in order to cover the robot. Assuming a limited speed for the robot, the next window can be one with its centre at the last calculated position. Hence, the appropriate window size is determined by the speed of the robot. In our case, we work with the Khepera miniature robot [4], which at maximum speed (approximately 1m sec. can not move very far during the sample time (20 millisec. of the frame grabber. Therefore, we can use a quite small window size. As an example, let us assume that 1 pixel corresponds to approximately 0.5 0.5 cm, that the maximum speed of the robot ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III. Lecture Notes in Control and Information Sciences 200, pages 501--513, Heidelberg, 1994. Springer-Verlag.

....6.270 MIT kit to be used in a package delivery application. Introduction Hardware miniaturization offers the opportunity to bridge the gap between laboratory mobile robot designs and industrial applications like floor cleaning, surveillance, flexible part transportation, delivery, or harvesting [13]. Each of the applications in this extended context requires increased sensing capabilities and processing power. System architectures are beginning to reflect the increase in complexity by integrating multiple micro controllers and relying on distributed processing. Consequently, software ....

.... it operates ffl enable evolution towards more complex or more refined patterns of operation Although our test robot hardware does not allow the connection of additional extensions based on microcontrollers, we consider that the software architecture should aim at a multi controller network as in [13]. Moreover it should make such an architecture transparent to the programmer. A modular architecture organizes the knowledge, competences and activity of the system such that local changes, improvements or tuning do not affect the functioning of unchanged components. It offers a higher level of ....

Francesco Mondada, Edoardo Franzi, and Paolo Ienne. Mobile robot miniaturisation: a tool for investigation in control algorithms. Third International Symposium on Experimental Robotics, Kyoto, Japan, Oct 28-30, 1993.

....in a somewhat circular fashion around its body and allow the measurement of distances in a short range from about 1 to 4 cm. Khepera also carries a Motorola 68331 micro controller. Two motor wheels are controlled by classical PID controllers and their speeds can be independently manipulated [1]. In a first step omitted in this paper we designed an ordinary fuzzy controller responsible for local obstacle avoidance. Driven by this controller the robot permanently moves in its environment and, when approaching an obstacle, changes its course in order to avoid any collision. The use of a ....

F. Mondada, E. Franzi, and P. Ienne, "Mobile robot miniaturisation: a tool for investigation in control algorithms", 3. Int. Symposium on Experimental Robotics, Kyoto, Japan, 1993.

....mobile robot designed to keep an arena clear by picking up trash objects and releasing them outside the arena. The robot has to look for garbage , somehow grasp it, and take it out of the arena (see [8] The organism is a miniature mobile robot called Khepera, developed at E.P.F.L. in Lausanne ([6]) The robot is supported by two wheels that allow it to move in various directions by regulating the speed of each wheel. In addition, the robot is provided with a gripper module with two degrees of freedom. The two arms of the gripper can move in parallel through any angle from vertical to ....

Mondada, Francesco, E. Franzi and P. Ienne,"Mobile robot miniaturisation: a tool for investigation in control algorithms", Experimental Robotics III, Lecture Notes in Control and Information Sciences (T. Yoshikawa and F. Miyazaki eds.), Springer-Verlag (1994).

....controlling the behavior of organisms living in a (simulated) physical environment [2] We describe the results of simulations comparing the behavior of haploid and diploid populations in both fixed and changing environments. 2. Simulations The organism is a simulated mobile robot called Khepera [3,4]. The environment is a rectangular box of 6Ox35 cm and the robot is rewarded (in the evolutionary sense) for exploring as much as possible of this environment. The environment includes a circular food area of 2O mm of diameter that may be located in different positions in the environment. The ....

F. Mondada, E. Franzi and P. Ienne, "Mobile Robot Miniaturisation: a Tool for investigation in control algorithms", in Proc. of the 3rd Int. Symposium on Experimental Robotics, Kyoto, 1993.

....simulations in which the task of the evolving populations is to explore an environment and return, time to time, to a food area where individuals can reintegrate the energy consumed during the exploration. The organism is a miniature mobile robot called Khepera, developed at E.P.F.L. in Lausanne (Mondada et al. 1993). Khepera has a circular shape with a diameter of 55 mm. a height of 30 mm. and a weight of 70 g. The robot is supported by two wheels that allow it to move in various directions by separately regulating the speed of each wheel. The robot is also provided with eight infra red proximity sensors ....

Mondada, F., Franzi, E. & Ienne, P. (1993). Mobile Robot Miniaturisation: a Tool for investigation in control algorithms. In: Proceedings of the third International Symposium on Experimental Robotics, Kyoto, Japan.

....3 A Robot specific Simulator. With reference to the calculations made earlier, we would like to make a simulator by building a look up table of possible sensor and motor responses for the robot used in the exploration and homing experiments. The robot is the Khepera miniature mobile robot [26]. Figure 1 shows the Khepera robot, which has become widely used in the field of Evolutionary Robotics during the last couple of years. Its inventors have sold more than 300 copies of the robot to universities and individuals. It is an excellent tool for robotics experiments because of its small ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Experimental Robotics III. Lecture Notes in Control and Information Sciences 200, pages 501--513, Heidelberg, 1994. Springer-Verlag.

....5 Application to an autonomous mobile robot 5. 1 Description of the robot In order to test the validity of our method, we evolved populations of chromosomes generating networks able to control a Khepera like simulated mobile robot, keeping in mind a final application to the real robot Khepera [12, 13]. This robot resembles Braitenberg vehicles [2] It has two motors, each controlling one wheel, and 8 infra red proximity sensors able to detect obstacles at a distance of a few centimeters. 5.2 Evolutionary process Natural selection uses a binary fitness function (life or death) We chose, as a ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Third International Symposium on Experimental Robotics, Kyoto, Japan, October 1993.

....the robot during a few laps in the environment. Let us consider the case of a mobile robot that navigates by performing a wall following behavior in an environment like that described in Figure 3. The robot (see Figure 4) is a miniature mobile robot developed at E.P.F.L. in Lausanne, Switzerland (Mondada, Franzi, Ienne, 1993). It has a circular shape with a diameter of 55 mm, a height of 30 mm, and a weight of 70g. It is supported by two wheels and two small Teflon balls. The wheels are controlled by two DC motors with an incremental encoder (10 pulses per mm of advancement by the robot) The robot is provided with ....

Mondada, F., Franzi, E. & Ienne, P. (1993). Mobile Robot miniaturisation: A tool for investigation in control algorithms, In: Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Japan.

....can be efficient enough to devise a way of escaping from rather complex mazes. Fig. 19. Overall plan discovered in a maze. IV. Real Robot Implementation To reproduce the simulated experiments and results of the previous section, the MonaLysa architecture has been used with a Khepera robot ( 28] [38]) shown in Figure 20. The size of each side of the square environment used for such a purpose was set at 70cm. The Khepera robot is equipped with six IR frontal sensors (with two additional sensors in the back) that can detect an obstacle within a range of approximately 5 cm, and with two wheels ....

F. Mondada, E. Franzi and P. Ienne, "Mobile Robot Miniaturisation: A Tool for Investigation in Control Algorithms," in Proceedings of the 3rd International Symposium on Experimental Robotics, 1993.

....map implementation of the Q learning is more recent [5] We propose to study the use and discuss the interest of this implementation comparing to a multilayer perceptron implementation or more classical ones. Experiments are performed in the real world with the miniature robot Khepera [6]. Q learning Reinforcement learning synthesises a mapping function between situations and actions by minimising a reinforcement signal. Q learning algorithms store the expected reinforcement value associated to each situationaction pair. Three different functions are involved: memorisation, ....

....neuron, the four neighbors are also updated. The learning coefficient is 0.9 for the selected neuron and 0.5 for the neighborhood. During the learning, the influence on the neighbors decreases proportionally to the inverse number of iterations. Experimentations Khepera is a miniature robot [6] having a diameter of 6 cm and a weight of 86 g. Two wheels allow the robot to move around. Eight infra red sensors help the robot to perceive its environment. The detection range is between 5 and 2 cm. Sensor data are ten bits real between 0.0 (nothing in front) and 1.0 (obstacle nearby) ....

Mondada F., Franzi E. & Ienne P., "Mobile Robot Miniaturisation: A Tool for Investigation in Control Algorithms," Third International Symposium on Experimental Robotics, Kyoto, Japan, October 1993.

....requirements so that a rather simple mobile robot can be used. Because no real robot was available, a physically realistic simulator had to be used. In particular, we have chosen the Khepera TM Simulator by Michel [28] The Khepera TM is a miniature mobile robot developed by Mondada et al. [30]. It has a circular shape with a diameter of 55 mm. The robot is built according to a modular concept as shown in Figure 4. In our experiments, the robot was used in the basic configuration, without vision and gripper. The Khepera TM can detect obstacles and light sources using its sensors. It ....

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Proceedings of the Third International Symposium on Experimental Robotics, pages 501-- 513, Kyoto, Japan, 1993. Springer.

....may reduce the discrepancy between the simulated and the embodied conditions. The authors also showed that there is an optimal amount of noise to use and that such optimal value can be computed by comparing performances in simulated and embodied conditions. Nolfi, Miglino, and Parisi (described in [16]) presented some experiments with the miniature mobile robot called Khepera (see below) that performed an obstacle avoidance task. The authors developed a simulator based on empirical sampled sensor readings and claimed that this sampling procedure can reduce the performance gap between the ....

....recognition task using a sonar. Such networks were then tested on a Nomad 200 robot with a built in wall following behavior. 3. The Experimental Setup 3.1 The robot and the environment In our experiments we used Khepera (see figure 1) a miniature mobile robot developed at E.P.F.L. in Lausanne [16]. Khepera has a circular shape with a diameter of 55 mm. a height of 30 mm. and a weight of 70g. it is supported by two wheels and two small teflon balls. The wheels are driven by extremely accurate stepper motors under P.I.D. control and can move both forward or backward. The robot is provided ....

Mondada F., Franzi, E., & Ienne, P. (1993). Mobile Robot miniaturisation: A tool for investigation in control algorithms. In: Proceedings of the third International Symposium on Experimental Robotics, Kyoto, Japan.

....that should be able to find and pick up cylindrical objects and to release them outside an arena surrounded by walls. 2. Methodological issues The experiments we describe in the next Sections involve the miniature mobile robot Khepera (Figure 1) developed at E.P.F.L. in Lausanne, Switzerland (Mondada, Franzi, and Ienne, 1993). 3 It has a circular shape with a diameter of 55 mm, a height of 30 mm, and a weight of 70g. It is supported by two wheels and two small Teflon balls. The wheels are controlled by two DC motors with an incremental encoder (10 pulses per mm of advancement by the robot) and they can move in ....

Mondada, F., Franzi, E., and Ienne, P. (1993). Mobile Robot miniaturisation: A tool for investigation in control algorithms. In: Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Japan.

....between goals, can be simulated by simple artificial Neural Networks (NN) where no complex computation is performed. We will present a real development and simulations about a Khepera TM robot (fig. 1) and a simulated system named Prometheus. Figure 1: The Khepera TM robot developed at the LAMI [Mon93]. We use a novel neural architecture named PerAc (Perception Action) which is a systematic way to decompose the control of an autonomous robot in perception and action flows [Gau94b] We show that action simplifies the interpretation of perception: each action is a choice and conditions entirely ....

Mondada F., Franzi, Ed., Ienne P., Mobile Robot Miniaturisation: A Tool For Investigation In Control Algorithms, Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Oct. 28-30, 1993.

....activity (I) 2. The evolution of an ability to navigate by using the physical robot Floreano and Mondada (in press) developed neural controllers for autonomus agents that should perform a navigation task by using an evolutionary approach. The robot used was Khepera, a miniature mobile robot (Mondada, Franzi and Ienne, 1993). Khepera has a circular shape with a diameter of 55 mm, a height of 30 mm and a weight of 70g; it is supported by two wheels and two small teflon balls. The wheels are controlled by two DC motors with an incremental encoder (10 pulses per mm of advancement of the robot) and can move in both F= V ....

Mondada F., E. Franzi, P. Ienne. 1993. Mobile Robot miniaturisation: A tool for investigation in control algorithms. In: Proceedings of the third International Symposium on Experimental Robotics, Kyoto, Japan.

No context found.

F. Mondada, E. Franzi, and P. Ienne, "Mobile robot miniaturisation: A tool for investigation in control algorithms," in Proc. of the 3rd Int. Symp. on Experimental Robotics. Heidelberg: Springer-Verlag, 1994, pp. 501--513.

No context found.

F. Mondada, E. Franzi, and P. Ienne. Mobile Robot Miniaturisation: A Tool for Investigation in Control Algorithms. In Proceedings of the Third International Symposium on Experimental Robotics, pages 501-- 513, Kyoto, 1993.

No context found.

Francesco Mondada, Edoardo Franzi, and Paolo Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. Proceedings of the Third International Symposium on Experimental Robotics, pages 501--513, 1993. 9

No context found.

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In T. Yoshikawa and F. Miyazaki, editors, Proceedings of the Third International Symposium on Experimental Robotics, pages 501--513, 1993.

No context found.

F. Mondada, E. Franzi, and P. Ienne. Mobile robot miniaturisation: A tool for investigation in control algorithms. In Proceedings of the Third International Symposium on Experimental Robotics, Kyoto, Japan, 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC