Bryson, J. 2000. The Study of Sequential and Hierarchical Organisation of Behaviour via Artificial Mechanisms of Action Selection. M.Phil Thesis. University of Edinburgh.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....running (see screenshot on next page) Unfortunately, the graphical interface only runs on machines that support a now obsolete window toolkit; if this were not the case, Tyrrell s simulated environment would be an excellent test bed for any newly proposed action selection mechanisms. However, Bryson [2000] integrated her mechanism into Tyrrell s environment, which fortunately can also be run without the graphical interface, to compare it with the best of the already implemented mechanisms, as discussed in section 2.3.3. 13 A screenshot from Tyrrell s Environment. The agent, a small rodent, is on ....

....that existing behaviours are correct when implementing a new behaviour. Edmund also attempts to combine the advantages of free flow hierarchical control, to keep the problem manageable, and explicit sequencing, to encode fixed action patterns, which can be followed over several time steps. In [Bryson, 00] Edmund was compared to the extended Rosenblatt and Payton architecture implemented by Tyrrell. Edmund was implemented in the small animal in Tyrrell s environment, and a large range of tests was performed. Edmund outperformed the extended Rosenblatt and Payton architecture by a small factor, ....

[Article contains additional citation context not shown here]

Bryson, J. 2000. The Study of Sequential and Hierarchical Organisation of Behaviour via Artificial Mechanisms of Action Selection. M.Phil Thesis. University of Edinburgh.

....perception rich behaviors Brooks originally described than are those reduced to being primitive function calls. Edmund behaviors and actionselection scripts can be implemented using a modified version of object oriented design (e.g. Coad et al. 1997) called behavior oriented design, or bod (Bryson, 2000a) The behaviors of an Edmund agent are responsible for monitoring the parts of the environment that are salient to them, for remembering relevant experiences (such as the agent s own current motivation or previous social encounters, and for controlling expression of their associated actions. ....

....both approaches to be particularly well represented because their agents in the se were designed by the same people who designed the architectures. We have discussed at length elsewhere the validity of testing action selection mechanisms in a simulated environment rather than on a robot (see Bryson, 2000b, Section 4.5) Although purpose built simulations can include biases which overlook significant issues, the same can be said of many robot experiments. In this case, the simulation has been well established in the literature, and has been previously demonstrated with a large variety of ....

Bryson, J. (2000b). The study of sequential and hierarchical organisation of behaviour via artificial mechanisms of action selection. MPhil Thesis, University of Edinburgh.

.... Modularity Work: Combining Memory Systems and Intelligent Processes in a Dialog Agent Joanna Bryson MIT AI Lab; Cambridge MA 02139; USA joanna ai.mit.edu Abstract One of the greatest obstacles to designing a mind is the complexity of integrating different process types, time frames and representational structures. This paper describes a methodology for addressing this obstacle, Behavior ....

Joanna Bryson. The study of sequential and hierarchical organisation of behaviour via artificial mechanisms of action selection. MPhil Thesis, University of Edinburgh, 2000.

....perception rich behaviors Brooks originally described than are those reduced to being primitive function calls. Edmund behaviors and actionselection scripts can be implemented using a modi ed version of object oriented design (e.g. Coad et al. 1997) called behavior oriented design, or bod (Bryson, 2000a) The behaviors of an Edmund agent are responsible for monitoring the parts of the environment that are salient to them, for remembering relevant experiences (such as the agent s own current motivation or previous social encounters, and for controlling expression of their associated actions. ....

....both approaches to be particularly well represented because their agents in the se were designed by the same people who designed the architectures. We have discussed at length elsewhere the validity of testing action selection mechanisms in a simulated environment rather than on a robot (see Bryson, 2000b, Section 4.5) Although purpose built simulations can include biases which overlook signi cant issues, the same can be said of many robot experiments. In this case, the simulation has been well established in the literature, and has been previously demonstrated with a large variety of ....

Bryson, J. (2000b). The study of sequential and hierarchical organisation of behaviour via articial mechanisms of action selection. MPhil Thesis, University of Edinburgh.

....autonomous agents (CAA) This assistance comes both directly, by providing understanding and support for the requirements of such artificial systems, and indirectly, by providing another AI model of biological intelligence . effectively requires decomposition into modules and hierarchy (see [10] for discussion and references. This requirement is not theoretical, but practical. In theory, monolithic systems may be Turing complete; but whether attacked by design or by learning, in practice complex control requires decomposition into solvable subproblems. This principle of modularity is ....

Joanna Bryson. The study of sequential and hierarchical organisation of behaviour via artificial mechanisms of action selection, 2000. MPhil Thesis, University of Edinburgh.

.... to theoretical opposition: Is a system truly autonomous if it is forced to carry out a plan Is centralized control biologically plausible The answer to both of these questions is almost certainly yes , but space does not permit the arguments here see for example Barber and Martin (1999) and Bryson 4 DISCUSSION AND RECOMMENDATIONS 22 (2000) respectively for some discussion. Regardless, it can be observed empirically that all autonomous agents do still have and require action selection mechanisms. In behavior based systems, these systems are often distributed across the behaviors. This may lead to some improvement of robustness, but ....

Bryson, J. (2000). The study of sequential and hierarchical organisation of behaviour via artificial mechanisms of action selection. M.Phil. thesis.

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