D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics--Part B: Cybernetics, 26(3), pages 396--407, 1996. 75 76 REFERENCES  Home/Search   Document Details and Download   Summary   Related Articles   Check

....scale to many control parameters, as then the search spaces can become unreasonably large. As described earlier, genetic algorithms 19 require much performance data to function, and even though other investigators have interleaved simulated and real world performance evaluation with some success [29], 60] accurate simulation is required to produce working controllers (which becomes more di#cult as task complexity increases) Also, control heterogeneity was permitted in all cases, which meant that public policies su#ered severely from the state space size problem, as parameter values for ....

....systems depend heavily on sensitive agent to agent and agent to environment interactions, performance is often stochastic, hence evaluative, rather than gradient based, search methods are appropriate. This type of control optimization has been extensively studied for the case of a single agent [93, 29, 111], as well as for multiple agents [82, 70] This design methodology assumes that all agents follow the same control policy and that the only metric used for system evaluation is the global task metric. The use of homogeneous controllers with a global reward signal provides a way of addressing the ....

D. Floreano and Mondada F. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man and Cybernetics, 26:396--407, June 1996.

....to this end. Here we brie y review a few of these studies, each demonstrating the use of a di erent technique. The activity of the 13 internal (hidden layer) neurons in the network as a function of a robot s location and orientation was charted via a simple form of receptive eld measurement in [65]. The function of these intermediate neurons was generally highly distributed and dependent on previous states, but a certain interneuron playing a major role in path planning was also identi ed. Others have systematically clamped neuronal activity and studied its e ects on the robot s behavior ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics - Part B, 26(3):396-407, 1996.

....have a common temporal behavior, making a transition from one quasi stable state of ring to another together. Dynamical modules give new insights to CPG operation, describing them in terms of a nite state machine, and enabling a rigorous analysis of their robustness to parameter variations. In [10], the activity of internal neurocontroller neurons as a function of a robot s location and orientation was charted by a simple form of receptive eld measurement. Neuronal functioning was generally highly distributed, but a speci c interneuron that had an important role in path planning was also ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics - Part B, 26(3):396-407, 1996.

.... and parameterized version of this model in order to evolve optimal synaptic learning rules for RL (with respect to maximizing nectar intake) using a genetic algorithm [26] In contrast to common evolutionary computation applications that involve NNs with evolvable synaptic weights or architectures [1, 10, 32], we set upon the task of evolving the network s neuronal learning rules. Previous attempts at evolving neuronal learning rules have used heavily constrained network dynamics and very limited sets of learning rules [4, 6, 11, 12] or evolved only a subset of the learning rule parameters [45] We ....

.... between evolving traits (such as synaptic weights) versus learning them [1, 19] A relatively small amount of research has been devoted to the evolution of the learning process itself, most of which was constrained to choosing the appropriate learning rule from a limited set of prede ned rules [4, 6, 10]. In this work we show for the rst time that optimal learning rules for RL in a general class of armed bandit situations, can be evolved in a general Hebbian learning framework. The evolved learning rules are by no means trivial, as they are heterosynaptic and employ synaptic plasticity ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man and Cybernetics, 26(3):396-407, 1996.

....systems depend heavily on sensitive agent toagent and agent to environment interactions, performance is often stochastic, and evaluative, rather than gradient based, search methods are appropriate. This type of control optimization has been extensively studied for the case of a single agent [14] [15], 16] as well as for multiple agents [17] 18] 19] II. The Flocking Task A. Task Definition The flocking task examined in this paper is similar in form to the cooperative movement task studied in [20] The agents begin each trial at random positions and orientations within an area A ....

D. Floreano and Mondada F., "Evolution of homing navigation in a real mobile robot," IEEE Transactions on Systems, Man and Cybernetics, vol. 26, pp. 396--407, June 1996.

....conducted. Much of this work was done using computer based simulations only [1] 2] 6] Examples of ER research conducted with real robots include the evolution of walking behaviors in hexapod and octopod robots [7] 8] and the evolution of simple behavioral controllers for small mobile robots[9][10]. The later include the development of phototaxis behaviors [11] 12] and of simple object avoidance [10] and navigation [13] For recent reviews of the field of ER see [14] 15] 16] 13] 1.2 Intelligence performance metrics in ER An ER application may make use of one or more types of fitness or ....

.... research conducted with real robots include the evolution of walking behaviors in hexapod and octopod robots [7] 8] and the evolution of simple behavioral controllers for small mobile robots[9] 10] The later include the development of phototaxis behaviors [11] 12] and of simple object avoidance [10] and navigation [13] For recent reviews of the field of ER see [14] 15] 16] 13] 1.2 Intelligence performance metrics in ER An ER application may make use of one or more types of fitness or intelligence metrics. These include 1) measurement of behavior quality for selection during training, 2) ....

D. Floreano and F. Mondada , Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, Cybernetics Part B: Cybernetics, Vol. 26, No. 3, pp. 396-407, 1996.

....Experimental results show that this approach has potential to develop a sophisticated neural controller for complex environments. 1. Introduction There are many studies of constructing mobile robot controller with different approaches such as evolving neural network by genetic algorithm [1], using genetic programming [2] combining fuzzy controller with genetic algorithm [3] and programming [4] In previous work [5] we presented CAM Brain, evolved neural networks based on cellular automata [5,6] and applied it to controlling a mobile robot. However, the controller composed of one ....

D. Floreano and F. Mortdana, "Evolution of homing navigation in a real mobile robot," [EEE Trans. Systems, Man, and Cybernetics, Vol. 26, No 3, pp. 396-407, $une, 1996.

....simulator. Simulation results show the possibility of easily developing higher behaviors by integrating CAM Brain behavior modules. 1. Introduction There are many studies of constructing mobile robot controller with different approaches such as evolving neural network by genetic algorithm [1], using genetic programming [2] combining fuzzy controller with genetic algorithm [3] and programming [4] In previous work [5] we presented CAM Brain, evolved neural networks based on cellular automata [5,6] and applied it to controlling a mobile robot. However, the controller obtained had a ....

D. Floreano and F. Mondana, "Evolution of homing navigation in a real mobile robot," IEEE Trans. Systems, Man, and Cybernetics, Vol. 26, No 3, pp. 396-407, June, 1996.

.... object of several trial and error experiments until the choice of the one with better performance (which in some cases needs a little help from the human, in establishing good situations) Moreover, a need to shorten evolution time is also identi ed by researchers using only GAs (see for example [2]) Consequently to appreciate the di erence between the two approaches we must rst consider where to place the border of human design and what we gain with its displacement in either way. 3 As to the second question, the use of EA in engineering research is interesting for obtaining parameter ....

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 26(3):396-407, June 1996.

....differing environments and remain operational when non vital components fail or become damaged. Previously, it has been demonstrated that robots can automatically acquire behaviours via a variety of learning methods, see [1] 2] 3] 4] 5] for reinforcement learning (RL) methods and [6] [7], 8] 9] for classifier (or evolutionary) techniques) However, these methods only produce successful results where the input to output state space is small enough to be learnt in real time. Generally, this restricts most robot learning experiments to be performed with toy robots equipped with ....

D. Floreano and F. Mondada, Evolution of Homing Navigation in a Real Mobile Robot, IEEE Trnas. on Sys. Man and Cybernetics, Vol. 23, No. 5, 1996.

....based decision control, using the model presented in [2] Other solutions, such as evolutionary programming or arti cial neural nets, might have been used. But the evolution or learning times involved and the speci c problems they would pose, would delay conclusive results (see, for instance, [3]) 3.1 A Procedural Representation The proposed solution to represent how to get back home uses the constituent behaviour modules themselves. The robot memorizes which behaviours have been active from the moment it left home until it nds the light source. Thereafter the robot will revert those ....

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, 26(3):396-407, June 1996.

....used was very similar to the one shown in Figure 1, except that it had eight input units for the eight infrared proximity sensors, three hidden (and context) units, and two outputs directly controlling the two motors. The fitness function used in experiment 1, similar to the one used in [24], rewarded robots for moving forward as straight and as quickly as possible while avoiding obstacles. As exemplified in Figure 6, robot controllers evolved collision free obstacle avoidance behaviour in both environments (in less than 100 generations) This confirms that Meeden s basic ....

Floreano, D., and Mondada, F., (1996). Evolution of Homing Navigation in a Real Mobile Robot. IEEE Transactions on Systems, Man, and Cybernetics -- Part B: Cybernetics, 26(3), 396-407.

....evolutionary computation [4] to develop ANNs. Work in this eld comprises of the development of isolated ANNs, evolving to maximize a certain target function on one hand [8, 5] and the development of embedded ANNs, serving as the control mechanism for an autonomous agent, on the other hand [3, 7, 11]. In the latter case the agents perform certain behavioural tasks, and their performance level in these tasks serves as the basis for evolutionary selection. This new paradigm of Evolved ANNs (EANNs) is clearly very interesting from the applicative point of view, opening new horizons in the ....

....non trivial network structures that evolve in these agents, we demonstrate the existence of neurons whose functional repertoire strongly resembles that of command neurons and place cells known from biological models. The emergence of place cells was already described by Floreano and Mondada [3], who study homing navigation of a real robot. We show that the evolution of such place cells in a task requiring simple navigation skills is a robust phenomenon, occurring under various sensory capabilities. Another important result presented here is the emergence of a memory mechanism. A ....

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics - Part B, 26(3):396-407, June 1996.

....possible because of minimal human design involvement in the product. It is not feasible that controllers for complete structures could be evolved (in simulation or otherwise) without first evolving controllers for simpler constructions. Compared to the traditional form of evolutionary robotics [5, 9, 13, 22, 26] which serially downloads controllers into a given piece of hardware, it is relatively easy to explore the space of body constructions in simulation. Realistic simulation is also crucial for providing a rich and nonlinear universe. However, 4 while simulation creates the ability to explore the ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, Part B, 26:396--407, 1996. 14

....selection procedure. Simulation results show the possibility of the action selection method for higher behaviors with CAMBrain modules. 1. Introduction There are many studies of constructing mobile robot controller with different approaches such as evolving neural network by genetic algorithm [1], using genetic programming, combining fuzzy controller with genetic algorithm [2] and programming. In previous work [3] we presented CAM Brain, evolved neural networks based on cellular automata [3,4] and applied it to controlling a mobile robot. However, the controller composed of one module ....

D. Floreano and F. Mondana, "Evolution of homing navigation in a real mobile robot," IEEE Trans. Systems, Man, and Cybernetics, vol. 26, no. 3, pp. 396-407, June, 1996.

.... and Rosch, 1991; Brooks, 1991; HendriksJansen, 1996; Clark, 1997; Pfeifer and Scheier, 1999) An attractive way to study systems that are embodied and situated is evolutionary robotics, i.e. the attempt to develop robots through a self organized process based on artificial evolution (Nolfi and Floreano, 2000). This approach stresses the importance of study system that have a real body, are situated in a physical environment, and develop their own skills in close interaction with the environment without human intervention. Current research in evolutionary robotics often involves agents that are mostly ....

Floreano D. and Mondada F. (1996a) Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics--Part B: Cybernetics, 26(3):396- 407.

....we use a generalized and parameterized version of this model in order to evolve optimal synaptic learning rules for RL (with respect to maximizing nectar intake) using a genetic algorithm. In contrast to common Alife applications which involve NNs with evolvable synaptic weights or architectures [4, 5, 15], we set upon the task of evolving the network s neuronal learning rules. Previous attempts at evolving neuronal learning rules have used heavily constrained network dynamics and very limited sets of learning rules [7, 9] We de ne a general framework for evolving learning rules, which essentially ....

.... exists or not; 6 real valued genes (range [ 1,1] specify the initial weights of the regular and di erential module synapses (the synaptic weight of the reward module is clamped to 1, e ectively scaling the other synapses) and two realvalued genes specify the action function parameters m (range [5,45]) and b (range [0,5] Thirteen remaining genes specify the learning dynamics: The regular and di erential modules each have a learning rule speci ed by 4 real valued genes (parameters A D of equation (3) range [ 1,1] The global learning rate is speci ed by a real valued gene; and four ....

[Article contains additional citation context not shown here]

Floreano D. and Mondada F. Evolution of homing navigation in a real mobile robot. IEEE Trans. on Systems, Man and Cybernetics, 26(3):396-407, 1996.

....argued [13] that the automated evolution by means of evolutionary algorithms is more appropriate than design by hand, since evolutionary algorithms are believed to allow for the development of complex control structures that exhibit many interactions between sub systems. Many research projects [6,8,9,13,19] have used genetic algorithms (GAs) to evolve neural control structures for autonomous agents. Until now, most research projects have focused on GAs, which are but one option of evolutionary algorithms. Evolutionary algorithms are a framework that comprises many di erent algorithms, such as ....

....results. Evidence will suggest that ESs are quite well suited for the research presented in this paper. Sections 4 and 5 show that a (3,6) ES, for example, evolves reasonable Braitenberg controllers within 30 generations, which takes approximately one and a half hours. Compared to other research [8,9,14,19] that uses GAs for an almost identical task, ESs accelerate the developmental process by more than an order of magnitude. This speed up is of great practical relevance, since the experiments are to be done with physical robots that dynamically interact with the real world. Also, such a time ....

[Article contains additional citation context not shown here]

D. Floreano, and F. Mondada, Evolution of Homing Navigation in a Real Mobile Robot, in: IEEE Transactions on Systems, Man, and Cybernetics-Part B , 26(3), (1996) 396-407.

....decomposition [6, 8] accessibility graph technique [7] incremental planning [12] probabilistic approach [17] potential field approach [15, 16, 1] and others. Moreover, different learning techniques have also been used by researchers to improve the performance of conventional controllers [4, 5]. Each of these methods has its own inherent limitations and is capable of solving only a particular type of problems. Canny and Reif [3] studied the computational complexity of some of these methods and showed that motion planning for a point robot in a two dimensional plane with a bounded ....

Floreano D. and Mondada F., Evolution of Homing Navigation in a Real Mobile Robot, IEEE Trans. on Systems, Man and Cybernetics-Part B: Cybernetics, 26(3), 396-407, 1996.

....navigation for an autonomous underwater vehicle (AUV) as well as shepherding [Schultz 96] and tracking for other mobile robots. Sammut 92] demonstrates machine learning of a reactive strategy to control a dynamic system by observing a controller that is already skilled in the task. While [Floreano 96] discusses similar work, the evolutionary process in this study is carried entirely online on the physical robot. In parallel to research of techniques of evolution of function, similar research is being done in the area of evolution of form. Funes 97] applied evolutionary techniques to the ....

Floreano, D. and F. Mondada. Evolution of Homing Navigation in a Real Mobile Robot. In IEEE Transactions on System, Man, and Cybernetics -- Part B; 26(3) 396-407; 1996.

....architecture for autonomous agents that combines continuous learning with predefined abilities. Robotics studies tend often, when developing a learning model, to address a particular problem of sensor actuator coordination, e.g. maze travelling ( 30] 39] spatial navigation and exploration ([13], 14] 25] or object manipulation ( 1] 34] where often only one direction of control (from sensor to actuator) is considered. In contrast, our approach tries to develop a single control architecture which enables a robot to learn and act independently of a specific task, environment or ....

....update rules and a winner take all based neural activation function. We call it DRAMA for Dynamical Recurrent Associative Memory Architecture. Different algorithms have been used to enable learning in autonomous mobile agents, e.g. Reinforcement Learning ( 1] 29] 43] 44] Genetic Algorithms ([13], 31] 32] and more recently Artificial Neural Network (ANN) architectures ( 26] 30] 39] 45] 46] Neural network architectures are more interesting to us than RL and GA techniques as they require little knowledge of the task by not relying on the design of a good evaluation function for ....

Floreano, D. & Mondada, F. (1996), `Evolution of Homing Navigation in a Real Mobile Robot', IEEE Transactions on Systems, Man, and Cybernetics--Part B: Cybernetics, 26(3), pp.396-407.

....agents that combines continuous learning with predefined abilities. Robotics studies often tend, when developing a learning model, to address a particular problem of sensoractuator coordination, e.g. maze traveling (Owen Nehmzow, 1996; Tani et al. 1997) spatial navigation and exploration (Floreano Mondada, 1996; Gaussier et al. 1998; Kuipers, 1987) or object manipulation (Asada et al. 1997; Pfeifer Scheier, 1998) where often only one direction of control (from sensor to actuator) is considered. In contrast, our approach tries to develop a single control architecture which enables a robot to learn ....

....function. We call it DRAMA for Dynamical Recurrent Associative Memory Architecture. Different algorithms have been used to enable learning in autonomous mobile agents, e.g. Reinforcement Learning (Asada et al. 1997; Mataric, 1997; Wyatt et al. 1998; Yanco Stein, 1993) Genetic Algorithms (Floreano Mondada, 1996; Nolfi, 1997; Nordin Banzhaf, 1996) and more recently Artificial Neural Network (ANN) architectures (Kurz, 1996; Owen Nehmzow, 1996; Tani et al. 1997; Zimmer, 1996; Zrehen, 1995) Neural network architectures are more interesting to us than RL and GA techniques as they require little ....

Floreano, D. & Mondada, F. (1996). Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics (Part B: Cybernetics), 263, 396-407.

....been applied to robot control is given by Prabhu and Garg [16] Most of these works, however, have concentrated on simulated systems and therefore have not had to deal with the ambiguities and constraints of the real world. The primary exception to this has been in the area of navigation (e.g. [1, 2, 3]) In 1996, Hougen et al. [5] presented a new connectionist system designed for task learning on a real robot. In the current paper, we present an extension of this learning system to a significantly more difficult task. Learning responses is generally classified into supervised and unsupervised ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Trans. on Systems, Man, and Cybernetics, 26, Part B(3), 1996.

....of the robots depends mainly on their on board computational power and their energy supply. In many bio inspired single robot experiments, the robot is connected to a workstation through a cable, which supplies the required energy and supports intensive computing, such as learning algorithms [6]. With many robots using cables becomes impossible: they would become entangled. Two further robot features, not necessarily required in single robot experiments but essential in collective robotics, are the capability to explicitly communicate with and distinguish the other teammates from the ....

....that the integration of learning methods can contribute strongly to design a team of self programming robots in view of predefined task. In the last few years reinforcement learning and genetic algorithms have been used to produce adaptive behaviour in the context of single robot applications [8, 6]. In multiple robots applications, where fitness is measured at team level, robots are faced with the credit assignment problem, which means the problem of deciding to what extent their own behaviour has contributed to the team s overall score [9] Two ways for bypassing this problem have been ....

D. Floreano and Mondada F. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man and Cybernetics, 26:396--407, June 1996.

....[351] Comput. Ind. Eng. UK) 229] Control Engineering Practice, 90] IEE Colloq. Dig. 262] IEE Conf. Publ. ETSI konferenssi, 265] IEEE Transactions on Evolutionary Computation, 301] IEEE Transactions on Industrial Electronics, 244] IEEE Transactions on Systems, Man, and Cybernetics, [258, 264, 270, 278, 324, 325, 348] IEICE Transactions, 435] IEICE Transactions on Information and Systems, 408] Information Sciences, 311] International Journal of Vehicle Design, 290] J. Intell. Robot. Syst. Theory Appl. Netherlands) 246] J. Jpn. Soc. Precision Eng. Japan) 308] J. Robot. Syst. USA) 99, 259, ....

....[27] Emmanuel, T. 177, 185] Enns, Russell, 168] 14 Genetic algorithms in robotics Erbudak, M. 37] Erkmen, A. M. 37] Espenschied, Kenneth S. 216] Fagg, Andrew H. 401] Falkenauer, Emanuel, 43, 44] Farritor, Shane, 263] Feng, S. 400] Fennel, Theron Randy, 16] Floreano, Dario, [110, 134, 181, 124, 264, 293] Fogarty, Terence C. 129, 180, 212] Fogel, David B. 349, 350, 351] Fogel, Lawrence J. 349, 350] Ford, Gary P. 45] Fraser, A. P. 68, 71, 88] Fu, Li Chen, 39] Fujimoto, Hideo, 21, 36] Fujimoto, S. 243] Fujimoto, Shinsaku, 155] Fukuda, T. 173, 191, 275, 294] Fukuda, Toshio, 69, ....

[Article contains additional citation context not shown here]

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, 26(3):396--407, 1996. ga96bFloreano.

....with real robots. In these cases the behaviors evolved have been simple and the robots have had few degrees of freedom with few sensors. Examples of autonomously evolved behaviors are: forward, backward and stopping behaviors with a wheeled robot in [10] homing navigation with a Khepera in [11]; and pursuer evader behaviors with Kheperas in [12] None of these behaviors would be particularly difficult to implement by hand nor would they be difficult to evolve in simulation (comparable behaviors have been successfully transferred from simulation to physical robot in [13] In our case ....

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transaction on Systems, Man and Cybernetics--Part B: Cybernetics, 26(3):396--407, 1996.

.... IEEE Transactions on Power Systems, 89, 116, 129, 153, 161, 206, 228, 233, 247, 283, 343, 667] IEEE Transactions on Reliability, 88] IEEE Transactions on Semiconductor Manufacturing, 397, 854] IEEE Transactions on Signal Processing, 712] IEEE Transactions on Systems, Man, and Cybernetics, [112, 252, 594, 624, 643, 649, 742, 743, 797, 867] IEEE Transactions on Systems, Man, and Cybernetics Part B: Cybernetics, 248, 391, 505] IEEE Transactions on Systems, Man, and Cybernetics A: Syst. Humans. 522] IEEE Transactions on Systems, Man, and Cybernetics B, Cybernetics, 822, 858] IEEE Transactions on Systems, Man, and ....

....Anne, 104] Ferland, Jacques A. 648, 824] Filipic, Bogdan, 123] Filman, D. J. 430] Finley, Linda, 539] Fischer, D. 21] Fisher, B. J. 39] Flasse, St ephane P. 391] Flasse, Stephane P. 767] Fleming, Peter J. 570] Fleming, Peter, 84] Fleurent, Charles, 648, 824] Floreano, Dario, [649] Flower, Joe, 124] Flowers, Woodie C. 530] Fogarty, Terence C. 816] Authors 23 Fogarty, Terence, 480] Fogel, David B. 650, 868] Fogel, Lawrence J. 868] Fonteix, C. 520] Forrest, Stephanie, 416, 651] Fortuna, L. 610] Fotouhi, Frashad, 575] Fraga, E. S. 125] Francone, ....

[Article contains additional citation context not shown here]

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, 26(3):396--407, 1996. ga96bFloreano.

....possible because of minimal human design involvement in the product. It is not feasible that controllers for complete structures could be evolved (in simulation or otherwise) without rst evolving controllers for simpler constructions. Compared to the traditional form of evolutionary robotics [15 19] which serially downloads controllers into a piece of hardware, it is relatively easy to explore the space of body constructions in simulation. Realistic simulation is also crucial for providing a rich and nonlinear universe. However, while simulation creates the ability to explore the space of ....

Floreano, D., Mondada, F.: Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics (1996)

.... end, we implemented a standard genetic algorithm using, for the assessment of individuals, a tness function measured via the robotic implementation of the phonotaxis mechanism described in [3] The experiments therefore used on line evolution, somewhat similar to the work of Floreano and Mondada [6, 7], except that the goal here was not to evolve robot controllers with particular capabilities but to follow a chosen evolutionary scenario the latter predetermined by the target biological system. The paper continues with a description of the speci c evolutionary scenario we chose to model. ....

Floreano, D., Mondada, F.: Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics 26 (1996) 396-407 Part B: Cybernetics.

....has been found to be implicit if the environment is cluttered enough, because the robot is obliged to avoid obstacles if it has to go as fast and as straight as possible. Also the D term in equation (1) is changed to the absolute value of the sum of the signed differences between wheel speeds. In [11], 42] 43] similar behaviors are evolved endowing the Khepera robot with a simulated metabolism such as, when the robot moves away from its initial position, its energy level increases and, conversely, when it moves towards the initial position, its energy level decreases. The robot is assumed ....

....zero. Its fitness value is the maximum distance it occurred to be from its initial position during its life time. Likewise, in other realizations, the robot is endowed with a simulated motivational system and different behaviors are sought depending upon which motivation is currently the highest ([11], 40] When the robot is equipped with the proper actuators, more elaborated behaviors like area cleaning can be evolved ( 54] 49] 50] 51] Sometimes also, besides controlling simple sensorimotor behaviors, neural controllers integrate perceptions and actions over time into some form ....

[Article contains additional citation context not shown here]

Floreano, D. and Mondada, F. Evolution of Homing Navigation in a Real Mobile Robot. IEEE Transactions on Systems, Man, and Cybernetics - Part B: Cybernetics. 26, 3, 396-407. 1996.

....of both approaches. Using the Voronoi diagram as an intermediate tool, the system extracts a topological map from a grid based model previously constructed by exploring the environment. Evolutionary learning approaches for mobile robot navigation, have also been proposed and experimented, e.g. [14]. A difficulty with these approaches is that learning a solution is considerably time consuming. To gain advantages from different approaches, this paper integrates ideas from multiresolution partition methods, and geometric primitive based and graph based methods. Specifically we demonstrate the ....

D. Floreano and F. Mondada, "Evolution of homing navigation in a real mobile robot," IEEE Trans. Syst., Man, Cybern. B, vol. 26, pp. 396--407, June 1996.

....Nagahama et al. 1994] The command neuron s activity is driven either by positional information or by a short term memory mechanism, depending on the speci c sensory information available to the agent. The emergence of location dependent cells in EANNs has previously been demonstrated 1 by [Floreano and Mondada, 1996] who studied homing navigation of a real robot. They found a neuron in the controlling EANN that exhibits location and orientation dependent activity. Jakobi, 1998] has described a simple memory based behaviour in a small, bilaterally symmetrical EANN with 10 neurons and about 20 synapses. In ....

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics - Part B, 26(3):396-407, June 1996.

.... empirically demonstrated that our method of evolving adaptive controllers can solve a complex sequential task involving multiple behaviors whereas evolution of genetically determined controllers fail to do so satisfactorily (Floreano Urzelai, 1999) Finally, in a very recent paper (Urzelai Floreano, 2000) we have shown that evolutionary adaptive controllers can adapt to environmental changes that involve new sensory characteristics and new spatial relationships of the environment. In this paper, we describe a new set of experiments designed to further test the robustness of this approach to ....

Floreano, D., & Mondada, F. (1996a). Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics-Part B, 26, 396--407.

....and of the robot hardware. The methodology has been applied to several types of robots, with wheels and legs (figure 2) When placed in more complex environments, robots can evolve neural mechanisms that build internal representations of space and time in relationship to internally defined goals [2]. When co evolved with a competing robot (figure 3) the reciprocal bootstrapping of the competing controllers drives the ecosystem to increased levels of com noback.eps figs 48 Theta 36 mm Figure 3: Prey and predator robots co evolved in competition with each other. The predator on the ....

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics-Part B, 26:396--407, 1996.

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics--Part B: Cybernetics, 26(3), pages 396--407, 1996. 75 76 REFERENCES

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, 1996.

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics---Part B: Cybernetics, 26(3):396--407, 1996.

No context found.

D. Floreano and F. Mondada, Evolution of Homing Navigation in a Real Mobile Robot, IEEE Transactions on Systems, Man, and Cybernetics-Part B, 26:396---407, 1996

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics-- Part B: Cybernetics, 3(26):396--407, 1996.

No context found.

Floreano, D. and Mondada, F., Evolution of Homing Navigation in a Real Mobile Robot, in IEEE Transactions on Systems, Man, and Cybernetics - Part B: Cybernetics, v. 26, n. 3, pp. 396-407, 1996.

No context found.

Dario Floreano and Francesco Mondada. Evolution of homing navigation in a real mobile robot. In [16], pages 402--4101, 1996.

No context found.

D. Floreano, F. Mondada, Evolution of homing navigation in a real mobile robot, IEEE Transactions on Systems Man, and Cybernetics. Part B: Cybernetics 26 (3) (1996) 396--407.

No context found.

D. Floreano, F. Mondada , Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, Cybernetics Part B: Cybernetics Vol. 26 No. 3, pp. 396-407, 1996.

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, 1996.

No context found.

D. Floreano, F. Mondada , "Evolution of homing navigation in a real mobile robot," IEEE Transactions on Systems, Man, Cybernetics Part B: Cybernetics, vol. 26 no. 3, pp. 396-407, 1996.

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions on Systems, Man, and Cybernetics, 1996.

No context found.

D. Floreano and F. Mondada. Evolution of homing navigation in a real mobile robot. IEEE Transactions of Systems, Man, and Cybernetics-Part B: Cybernetics, 26(3):396-- 407, 1996.

No context found.

,

No context found.

D. Floreano, and F. Mondada, "Evolution of Homing Navigation in a Real Mobile Robot", IEEE Transactions on Systems, Man and Cybernetics, 26(3):396--407, 1996.

No context found.

D. Floreano and F. Mondada, "Evolution of homing navigation in a real mobile robot," IEEE Transactions on System Man and Cybernetics 26(3), pp. 396--407, 1996.

First 50 documents  Next 50

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