The advent of deep learning has inspired research into end-to-end learning for a variety of problem domains in robotics. For navigation, the resulting methods may not have the generalization properties desired let alone match the performance of traditional methods. Instead of learning a navigation policy, we explore learning an adaptive policy in the parameter space of an existing navigation module. Having adaptive parameters provides the navigation module with a family of policies that can be dynamically reconfigured based on the local scene structure, and addresses the common assertion in machine learning that engineered solutions are inflexible. Of the methods tested, reinforcement learning (RL) is shown to provide a significant performance boost to a modern navigation method through reduced sensitivity of its success rate to environmental clutter. The outcomes indicate that RL as a meta-policy learner, or dynamic parameter tuner, effectively robustifies algorithms sensitive to external, measurable nuisance factors.
I. IAutonomous navigation through static, unstructured environments has advanced in the past decades but fundamentally still relies on engineered approaches [1], [2]. Given an approximate map, the approaches use sensor data to inform updated estimates of the environment which are used to evaluate future trajectories in terms of safety and other characteristics, with the aim of finding a collision-free, goalattaining path. Traditionally designed systems involve manual parameter selection for general purpose navigation, which exhibits sensitivity to environmental conditions. This paper investigates the use of machine learning to dynamically reconfigure the parameters of a hierarchical navigation system according to the immediate, sensed surroundings of the robot. We show that scene-dependent online tuning improves navigation performance and reduces sensitivity to environmental conditions. The final reinforcement learning solution, called NavTuner, addresses the problem of parameter sensitivity to operational variance.
A. Research Context 1) Navigation and Machine Learning:One candidate approach to learning and navigation is to replace the traditionally engineered system with an end-to-end sensor to decision neural network [3]- [6]. Empirical and limited benchmarking show some promise on this front. However, instead of directly solving the navigation problem itself, these methods solve