Abstract
The increasing deployment of robots in co-working scenarios with humans has revealed complex safety and efficiency challenges in the computation of robot behavior. Movement among humans is one of the most fundamental—and yet critical—problems in this frontier. While several approaches have addressed this problem from a purely navigational point of view, the absence of a unified paradigm for communicating with humans limits their ability to prevent deadlocks and compute feasible solutions. This paper presents a joint communication and motion planning framework that selects from an arbitrary input set of robot's communication signals while computing robot motion plans. It models a human co-worker's imperfect perception of these communications using a noisy sensor model and facilitates the specification of a variety of social/workplace compliance priorities with a flexible cost function. Theoretical results and simulator-based empirical evaluations show that our approach efficiently computes motion plans and communication strategies that reduce conflicts between agents and resolve potential deadlocks.
Introduction
Navigating shared workspaces with humans remains a primary hurdle for autonomous systems. Traditional motion planners often treat humans as dynamic obstacles, leading to two critical failures: the "robot freezing problem" and intent ambiguity. In confined spaces, a robot may stop indefinitely because every potential movement appears to violate safety constraints. Furthermore, relying solely on implicit communication—the robot's physical trajectory—often fails to convey intent quickly enough to avoid conflict. This research proposes that robots must utilize explicit communication (e.g., digital signals, lights, or voice) to actively influence human behavior and facilitate mutually feasible navigation paths.
Methodology
The proposed framework, termed Deliberative Communication Planning (PDC), integrates two primary modules to synchronize physical action with intent signaling.
Motion Planner (MP)
The MP utilizes a modified CBF-TB-RRT (Control Barrier Function enabled Time-Based Rapidly-exploring Random Tree). This algorithm generates a library of diverse, collision-free path segments. Safety is mathematically guaranteed through Control Barrier Functions, which ensure the robot remains within a defined "safe set" of states relative to human co-workers.
Communication Planner (CP)
The CP evaluates combinations of motion plans and explicit communication signals (e.g., "Going North," "Staying Silent") using a search tree approach.
- Belief Modeling: The CP maintains a probabilistic model of the human's belief regarding the robot's intended path. A noisy sensor model accounts for human fallibility, such as misinterpreting or missing a signal.
- Joint Optimization: The system selects the (action, signal) pair that minimizes a multi-objective cost function. This function balances robot and human travel times, safety proximity, and the "cost" of communication.
Experimental Setup
The framework was evaluated in three custom simulation environments designed to test specific interaction challenges:
- Basic: A spacious map for general navigation.
- Hallway: A restricted environment for head-on encounters.
- Intersection: A confined space where only one agent can pass at a time, specifically designed to trigger deadlocks.
Human behavior was simulated using the Social Forces Model (SFM) and the Dynamic Window Approach (DWA).
Results and Discussion
The joint approach was compared against a baseline CBF-TB-RRT without explicit communication. The results demonstrate significant improvements across all metrics:
- Deadlock Resolution: While the baseline frequently resulted in deadlocks in the Intersection and Hallway maps, the joint planner resolved these conflicts 100% of the time.
- Efficiency and Safety: In the Basic environment, the joint planner reduced planning iterations from an average of 46–113 down to just 2. Proximity costs—measuring uncomfortable closeness—were dramatically lower, indicating safer and more decisive behavior.
- Flexibility: By tuning the cost function, the robot can prioritize its own efficiency or the human's comfort. When the robot was prioritized, its normalized speed increased by up to 2.7x compared to human-prioritized scenarios.
Conclusion
This work demonstrates that explicit communication is a powerful planning tool, not just a courtesy. By modeling human perception and intent, robots can "open" previously blocked paths and avoid the freezing problem. Future work will focus on scaling this reasoning to multi-agent environments and exploring diverse communication modalities like Augmented Reality and natural language.
Summary updated