Active Uncertainty Reduction for HRI: An Implicit Dual Control Approach

The ability to accurately predict human behavior is central to the safety and efficiency of robot autonomy in interactive settings. Unfortunately, robots often lack access to key information on which these predictions may hinge, such as people’s goals, attention, and willingness to cooperate.

In order to plan safely and efficiently, the robot must effectively reason about the human’s future motion and their reactions to the robot’s actions. Also, the robot needs to identify the human’s hidden states that are not observable by the robot but are important to the planning quality. Those hidden states might include the human’s objective, whether they are cooperative, or the role they prefer to play when interacting with the robot.

Dual control theory addresses this challenge by treating unknown parameters of a predictive model as stochastic hidden states and inferring their values at runtime using information gathered during system operation. While able to optimally and automatically trade off exploration and exploitation, dual control is computationally intractable for general interactive motion planning, mainly due to the fundamental coupling between robot trajectory optimization and human intent inference.

An example scenario tree that produces a dual control policy.

We propose a novel algorithmic approach to enable active uncertainty reduction for interactive motion planning based on the implicit dual control paradigm. Our approach relies on sampling-based approximation of stochastic dynamic programming, leading to a scenario-based stochastic model predictive control (SMPC) problem that can be readily solved by real-time gradient-based optimization methods

Our proposed ID-SMPC planner yielded a clean and sharp overtaking maneuver of the robot while the non-dual planners led to unsafe trajectories, due to lack of dual control effect. The ED-SMPC planner, even with fine tuning, was more conservative and unable to produce an efficient trajectory.

The resulting policy preserves the dual control effect for a broad class of predictive human models with both continuous and categorical uncertainty. We demonstrate our approach with simulated driving examples.

Simulation snapshots of a multi-agent interaction scenario with a pedestrian and two human-driven vehicles using the ID-SMPC planner.
%d bloggers like this: