UC Berkeley

Human-robot interactions (HRI) have been recognized to be a key element of future robots in many application domains such as manufacturing, transportation, service and entertainment. These applications entail huge social and economical impacts. Future robots are envisioned to function as human's counterparts, which are independent entities that make decisions for themselves; intelligent actuators that interact with the physical world; and involved observers that have rich senses and critical judgements. Most importantly, they are entitled social attributions to build relationships with humans. We call these robots co-robots.

Technically, it is challenging to design the behavior of co-robots. Unlike traditional robots that work in structured and deterministic environments, co-robots need to operate in highly unstructured and stochastic environments. The fundamental research question to address in this dissertation is how to ensure that co-robots operate efficiently and safely in dynamic uncertain environments.

The focus of this dissertation is 1) to set up a unified analytical framework for various human-robot systems; 2) to establish a methodology to design the robot behavior to address the fundamental problem.

A multi-agent framework to model human-robot systems is introduced in Chapter 2. In order to address the uncertainties during human-robot interactions, a unique parallel planning and control architecture is introduced in Chapter 2, which has a cognition module for human behavior estimation and human motion prediction, a long term global planner to ensure efficiency of robot behavior, and a short term local planner to ensure real time safety under uncertainties. The functionalities of these components are discussed in Chapter 3 to Chapter 5. Chapter 3 discusses the cognition module, which includes offline classification and online adaptation of various human behaviors. Chapter 4 and Chapter 5 discuss the optimal control or optimization problems for short term and long term robot motion planning. In a cluttered environment, the optimization problems are highly nonlinear and non-convex, hence hard to solve in real time, which may delay the robot's response in emergency situations. Fast online algorithms are developed to handle the issue: the convex feasible set algorithm (CFS) for the long term optimization, and the safe set algorithm (SSA) for the short term optimization. In particular, the CFS algorithm transforms the non-convex optimization problem into a sequence of convex optimization problems that can be solved efficiently online, which converges in fewer iterations and runs faster than conventional non-convex optimization solvers as shown in Chapter 6.

A method for theoretical evaluation of the designed behaviors is discussed in Chapter 7. The experimental platforms to evaluate the design are discussed in Appendix A. Applications of the proposed method on different co-robots are discussed in Chapter 8 and Chapter 9. Chapter 8 illustrates the application on automated vehicles in the framework of the robustly safe automated driving (ROAD) system. Chapter 9 discusses the application on industrial collaborative robots in the framework of the robot safe interaction system (RSIS).