Testing Generalization in Learned Simulated Robot Behaviors
نویسندگان
چکیده
Barriers to empirical comparisons of architectures are myriad and daunting, especially for physical robots. Architectures are implemented on different platforms that are often incompatible. Replicability is difficult to produce, and the multiple trials required for statistical significance are simply impractical or infeasible. The design criteria may be quite different as well, leading to potentially unfair conclusions that “my architecture is better than yours because it was designed to do something yours wasn’t”.
منابع مشابه
Generalization and Transfer Learning in Noise-Affected Robot Navigation Tasks
When a robot learns to solve a goal-directed navigation task with reinforcement learning, the acquired strategy can usually exclusively be applied to the task that has been learned. Knowledge transfer to other tasks and environments is a great challenge, and the transfer learning ability crucially depends on the chosen state space representation. This work shows how an agent-centered qualitativ...
متن کاملNeural Networks Robot Controller Trained with Evolution Strategies
Neural Networks (NN) can be used as controllers in autonomous robots. The specific features of the navigation problem in robotics make the generation of good training sets for the NN very difficult. In this paper an evolution strategy (ES) is introduced to learn the weights of the NN instead of the learning method of the network. The ES is used to learn high-performance reactive behavior for na...
متن کاملRobot Shaping: the Hamster Experiment
In this paper we present an example of the application of a technique, which we call robot shaping, to designing and building learning autonomous robots. Our autonomous robot (called HAMSTER1) is a multi-sensor mobile robot that performs the task of collecting “food” and bringing it to its “nest”. Its control architecture is based on the behavioral paradigm. The behavioral modules are implement...
متن کاملLearning High-level Behaviors from Demonstration through Semantic Networks
In this paper we present an approach for high-level behavior recognition and selection integrated with a low-level controller to help the robot to learn new skills from demonstrations. By means of Semantic Network as the core of the method, the robot gains the ability to model the world with concepts and relate them to low-level sensory-motor states. We also show how the generalization ability ...
متن کاملRoboShepherd: Learning a complex behavior
This paper reports on recent results using genetic algorithms to learn decision rules for complex robot behaviors. The method involves evaluating hypothetical rule sets on a simulator and applying simulated evolution to evolve more eeective rules. The main contributions of this paper are (1) the task learned is a complex behavior involving multiple mobile robots, and (2) the learned rules are v...
متن کامل