Abstract:Industrial robot manipulators are widely used in manufacturing processes. These robots are designed to carry out repetitive tasks. For safety concerns, they are often kept in cages, humans control them remotely via separate interfaces, physical Human-Robot Interaction (pHRI) is forbidden. Shortly all this is going to change radically. Next generation of the robots are expected to interact with humans directly. They will collaborate with people in applications such as assisted industrial manipulation, elder care, rehabilitation etc. To enable interaction between humans and robots, human safety has to be guaranteed. One of the pioneers in robot safety, Prof. Alessandro De Luca in his keynote speech at the International Conference on Robotics and Automation (ICRA) 2015, introduced the hierarchy of control architecture in pHRI which puts safety as the fundamental feature of a robot that has to work close to human beings (See Figure 1). Without safety, coexistence and collaboration can not take place. Collaboration Coexistence Safety Figure 1: Hierarchy of control architecture, reproduced with reference to [18] Current state of the art safety strategies include motion planning and collision detection and reaction. At high level control, motion planning is always preferred, as it produces natural movement of the robot. But once the planning fails to find a feasible trajectory to avoid collision, the robot is likely to collide with the object in the environment. Then it is crucial to detect the collision and react to it as soon as possible. This thesis focus on the augmentation of the current safety layer of the robot control structure by adding on a smart emergency braking layer. The braking is initiated by the prediction of the braking trajectory. The thesis proposed a trajectory prediction framework combined with the braking controllers. Two different braking controllers are tested in combination with the trajectory prediction scheme. Both results in collision avoidance of the robot in real-time simulation. One overall collision avoidance scheme has been tested successfully on the DLR Lightweight Robot III (LWR-III). In the test, the robot successfully initiated braking to avoid the collision.

The dynamic window approach to collision avoidance

Whole-body Dynamic Collision Avoidance with Time-varying Control Barrier Functions

Dynamical Obstacle Avoidance of Task- Constrained Mobile Manipulation Using Model Predictive Control

Dual-mode Dynamic Window Approach to Robot Navigation with Convergence Guarantees

Improvement of Dynamic Window Approach Using Reinforcement Learning in Dynamic Environments

Dynamic Adaptive Dynamic Window Approach

Reciprocal Collision Avoidance for Nonholonomic Mobile Robots

Dynamics-aware navigation among moving obstacles with application to ground and flying robots

An improved dynamic window approach algorithm for dynamic obstacle avoidance in mobile robot formation

Intelligent Optimization of Adaptive Dynamic Window Approach for Mobile Robot Motion Control Using Fuzzy Logic

Real-Time Navigation In Dynamic Human Environments Using Optimal Reciprocal Collision Avoidance

Safe and efficient collision avoidance control for autonomous vehicles

Reactive Collision Avoidance for Safe Agile Navigation

Improvement of Dynamic Window Approach in Dynamic Obstacle Environment

Proofs of the Technical Results Justifying an Algorithm for Collision Avoidance in Dynamic Environments with Moving and Deforming Obstacles

Finite Distribution Estimation-Based Dynamic Window Approach to Reliable Obstacle Avoidance of Mobile Robot

Biologically Inspired Collision Avoidance Without Distance Information

Real-time Collision Avoidance for Robot Manipulators by Braking Trajectory Prediction

Local path planning for autonomous mobile robots by integrating modified dynamic‐window approach and improved follow the gap method

On-line collision avoidance for collaborative robot manipulators by adjusting off-line generated paths: An industrial use case

Deadlock-Free Collision Avoidance for Nonholonomic Robots