Avoidance of Manual Labeling in Robotic Autonomous Navigation Through Multi-Sensory Semi-Supervised Learning.
Imitation learning holds the promise to address challenging robotic taskssuch as autonomous navigation. It however requires a human supervisor tooversee the training process and send correct control commands to robotswithout feedback, which is always prone to error and expensive. To minimizehuman involvement and avoid manual labeling of data in the robotic autonomousnavigation with imitation learning, this paper proposes a novel semi-supervisedimitation learning solution based on a multi-sensory design. This solutionincludes a suboptimal sensor policy based on sensor fusion to automaticallylabel states encountered by a robot to avoid human supervision during training.In addition, a recording policy is developed to throttle the adversarial affectof learning too much from the suboptimal sensor policy. This solution allowsthe robot to learn a navigation policy in a self-supervised manner. Withextensive experiments in indoor environments, this solution can achieve nearhuman performance in most of the tasks and even surpasses human performance incase of unexpected events such as hardware failures or human operation errors.To best of our knowledge, this is the first work that synthesizes sensor fusionand imitation learning to enable robotic autonomous navigation in the realworld without human supervision.
Continue reading and listening
Stay in the loop.
Subscribe to our newsletter for a weekly update on the latest podcast, news, events, and jobs postings.