Real-time obstacle avoidance for manipulators and mobile robots
International Journal of Robotics Research
Potential fields and neural networks
The handbook of brain theory and neural networks
Neural Networks in Robotics
Learning reactive and planning rules in a motivationally autonomousanimat
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Evolution of homing navigation in a real mobile robot
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Rapid, safe, and incremental learning of navigation strategies
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
An incremental approach to developing intelligent neural networkcontrollers for robots
IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics
Acquiring Mobile Robot Behaviors by Learning Trajectory Velocities
Autonomous Robots
Mobile Robot Learning by Self-Observation
Autonomous Robots
Mechanistic versus phenomenal embodiment: Can robot embodiment lead to strong AI?
Cognitive Systems Research
Hi-index | 0.00 |
The aim was to investigate a method of developing mobile robotcontrollers based on ideas about how plastic neural systems adapt totheir environment by extracting regularities from the amalgamatedbehavior of inflexible (nonplastic) innate subsystems interactingwith the world. Incremental bootstrapping of neural networkcontrollers was examined. The objective was twofold. First, todevelop and evaluate the use of prewired or innate robotcontrollers to bootstrap backpropagation learning for MultilayerPerceptron (MLP) controllers. Second, to develop and evaluate a newMLP controller trained on the back of another bootstrapped controller.The experimental hypothesis was that MLPs would improve on theperformance of controllers used to train them. The performances ofthe innate and bootstrapped MLP controllers were compared in eightexperiments on the tasks of avoiding obstacles and finding goals.Four quantitative measures were employed: the number of sensorimotorloops required to complete a task; the distance traveled; the meandistance from walls and obstacles; the smoothness of travel. Theoverall pattern of results from statistical analyses of thesequantities supported the hypothesis; the MLP controllers completed thetasks faster, smoother, and steered further from obstaclesand walls than their innate teachers. In particular, a single MLPcontroller incrementally bootstrapped by a MLP subsumption controllerwas superior to the others.