Next: Reinforcement Learning Up: Future Work Previous: Mapping and Navigation
One of the first things to explore in our coordination architecture is the use of a more economic view of the bidding mechanism. With this approach, each system (or agent) would be assigned a limited credit, and they would only be allowed to bid if they had enough credit. There should also be a way to reward the systems (agents). If not, they would run out of credit after some time and no one would be able to bid. The difficulty of the reward mechanism is how to decide when to give a reward and who deserves to receive it. This problem, known as the credit assignment problem, is very common in multiagent learning systems, especially in Reinforcement Learning, and there is not a general solution for it; each system uses an ad hoc solution for the task being learned.
An alternative to the economic view would be to have a mechanism to evaluate the bidding of each system (agent), assigning them succeeding or failing bids, or some measure of trust, in order to take or not take into account their opinions. However, we would face again the credit assignment problem.
Regarding the specific set of agents we have designed for solving the navigation problem, we could introduce some improvements on some of them, and even add new agents to the Navigation system. Some of these improvements could go in the following lines:
Some improvements could also be done on the Pilot and Vision systems. Regarding the Pilot, we could use a better obstacle avoidance algorithm. With the current algorithm, only the closest obstacle is considered for computing the avoidance path. We could improve the robot's performance if the Pilot took into account all the obstacles and landmarks stored in the Visual Memory, thus, producing better avoidance paths We are also planning to equip the robot with a laser scanner. This laser would be continuously scanning a 180 degree area in front of the robot to accurately detect obstacles that are several meters away. With this new sensor, the Pilot could avoid the obstacles before bumping into them, thus, generating better paths. Regarding the Vision system, we plan several improvements. The first one is to finish the stereo algorithm, so we can use the two available cameras. Another very important improvement is to make the Vision system more robust, so that it does not need to check the recognized landmarks against the Visual Memory. Actually, we should use the robust Vision system to adjust the imprecisions of the Visual Memory. We also plan to convert the Vision system into a Multiagent Vision system. In this system, several agents would process the camera images with different algorithms, and the agents should agree on what could be a good landmark (salient enough, robust, static, etc.). A final improvement of the Vision system would be to let it bid for services by other systems (either the Pilot system or itself). With the bidding capability, it could request the Pilot to approach a landmark to better recognize it, or even ``request itself'' to slightly move the camera so that a partially seen landmark enters completely the view field.
© 2003 Dídac Busquets