Highly Parallelized Data-Driven MPC for Minimal Intervention Shared Control

Abstract

We present a shared control paradigm that improves a user's ability to operate complex, dynamic systems in potentially dangerous environments without a priori knowledge of the user's objective. In this paradigm, the role of the autonomous partner is to improve the general safety of the system without constraining the user's ability to achieve unspecified behaviors. Our approach relies on a data-driven, model-based representation of the joint human-machine system to evaluate, in parallel, a significant number of potential inputs that the user may wish to provide. These samples are used to (1) predict the safety of the system over a receding horizon, and (2) minimize the influence of the autonomous partner. The resulting shared control algorithm maximizes the authority allocated to the human partner to improve their sense of agency, while improving safety. We evaluate the efficacy of our shared control algorithm with a human subjects study (n=20) conducted in two simulated environments: a balance bot and a race car. During the experiment, users are free to operate each system however they would like (i.e., there is no specified task) and are only asked to try to avoid unsafe regions of the state space. Using modern computational resources (i.e., GPUs) our approach is able to consider more than 10,000 potential trajectories at each time step in a control loop running at 100Hz for the balance bot and 60Hz for the race car. The results of the study show that our shared control paradigm improves system safety without knowledge of the user's goal, while maintaining high-levels of user satisfaction and low-levels of frustration. Our code is available online at https://github.com/asbroad/mpmi shared control.

Skill-based Shared Control

“…Recent literature introduces two different approaches to solve (1) for obstacle avoidance [6,27]. Both of these approaches minimize an objective function defined by cost(x, u; h)…”

“…Wu dt where x is the solution ofẋ = f (x, h) such that x(t c ) = x c is known. Broad et al [6] sample N ≫ 1 controls {u} N ∼ U, exclude samples such that x / ∈ X( e), and choose x, u with minimal cost. Rubagotti et al [27] use an off-the-shelf solver to compute a local minimizer.…”

“…Many shared control paradigms developed in the literature can be thought of as direct control within a restricted workspace. This property benefits a novice by only allowing them to operate within safe limits -a concept that lets the operator largely dictate motion, under the principle of minimal intervention [6,28]. While the principle achieves the crucial goal of minimally disrupting an operator, we advocate that shared control systems should, in addition, assist novice operators [9,11].…”

“…Other critical issues that the shared control literature addresses are, for instance, assisted guidance [26], human intention prediction [15,17], appropriate blending between human and autonomous policies [8], large time-delays [31], appropriate teleoperation spaces [22], obstacle avoidance [6,23,27], and representation and utilization of expert demonstrations [1]. However, these approaches tend to focus on cases where the intention of the operator consists of some distinct goal, e.g.…”

Data-driven Koopman operators for model-based shared control of human–machine systems

“…SP-DMD takes a large initial set of randomly generated basis functions and imposes an ℓ 1 penalty during the learning process to algorithmically decides which basis functions are the most relevant to the observable dynamics (Tibshirani, 1996). An example of this purely data-driven approach being applied to human–machine systems can be found in Broad et al (2019).…”