Learning Autonomous Mobility Using Real Demonstration Data
This work addresses the challenge of programming robust control laws for autonomous robot behaviors in uncertain real-world scenarios, such as slippage and obstacles, which is incremental as it builds on existing demonstration-based learning methods.
The authors tackled the problem of learning autonomous mobility for a tracked robot by developing a learning-based framework that uses human teleoperated demonstrations to learn feedback control policies, achieving tasks like obstacle negotiation and staircase traversal with successful policies learned from a few real demonstrations.
This work proposed an efficient learning-based framework to learn feedback control policies from human teleoperated demonstrations, which achieved obstacle negotiation, staircase traversal, slipping control and parcel delivery for a tracked robot. Due to uncertainties in real-world scenarios, eg obstacle and slippage, closed-loop feedback control plays an important role in improving robustness and resilience, but the control laws are difficult to program manually for achieving autonomous behaviours. We formulated an architecture based on a long-short-term-memory (LSTM) neural network, which effectively learn reactive control policies from human demonstrations. Using datasets from a few real demonstrations, our algorithm can directly learn successful policies, including obstacle-negotiation, stair-climbing and delivery, fall recovery and corrective control of slippage. We proposed decomposition of complex robot actions to reduce the difficulty of learning the long-term dependencies. Furthermore, we proposed a method to efficiently handle non-optimal demos and to learn new skills, since collecting enough demonstration can be time-consuming and sometimes very difficult on a real robotic system.