Supervised learning and reinforcement learning of feedback models for reactive behaviors: Tactile feedback testbed

Robots need to be able to adapt to unexpected changes in the environment such that they can autonomously succeed in their tasks. However, hand-designing feedback models for adaptation is tedious, if at all possible, making data-driven methods a promising alternative. In this paper, we introduce a full framework for learning feedback models for reactive motion planning. Our pipeline starts by segmenting demonstrations of a complete task into motion primitives via a semi-automated segmentation algorithm. Then, given additional demonstrations of successful adaptation behaviors, we learn initial feedback models through learning-from-demonstrations. In the final phase, a sample-efficient reinforcement learning algorithm fine-tunes these feedback models for novel task settings through few real system interactions. We evaluate our approach on a real anthropomorphic robot in learning a tactile feedback task.

An Efficient Online Trajectory Generation Method Based on Kinodynamic Path Search and Trajectory Optimization for Human-Robot Interaction Safety

With the rapid development of robot perception and planning technology, robots are gradually getting rid of fixed fences and working closely with humans in shared workspaces. The safety of human-robot coexistence has become critical. Traditional motion planning methods perform poorly in dynamic environments where obstacles motion is highly uncertain. In this paper, we propose an efficient online trajectory generation method to help manipulator autonomous planning in dynamic environments. Our approach starts with an efficient kinodynamic path search algorithm that considers the links constraints and finds a safe and feasible initial trajectory with minimal control effort and time. To increase the clearance between the trajectory and obstacles and improve the smoothness, a trajectory optimization method using the B-spline convex hull property is adopted to minimize the penalty of collision cost, smoothness, and dynamical feasibility. To avoid the collisions between the links and obstacles and the collisions of the links themselves, a constraint-relaxed links collision avoidance method is developed by solving a quadratic programming problem. Compared with the existing state-of-the-art planning method for dynamic environments and advanced trajectory optimization method, our method can generate a smoother, collision-free trajectory in less time with a higher success rate. Detailed simulation comparison experiments, as well as real-world experiments, are reported to verify the effectiveness of our method.