Constraint-aware estimation of human intent is essential for robots to physically collaborate and interact with humans. Further, to achieve fluid collaboration in dynamic tasks intent estimation should be achieved in real-time. We present a framework that combines online estimation and control to facilitate robots in interpreting human intentions, and dynamically adjusting their actions to assist in dynamic object co-manipulation tasks while considering both robot and human constraints. Central to our approach is the adoption of a Dynamic Systems (DS) model to represent human intent. Such a low-dimensional parameterized model, along with human manipulability and robot kinematic constraints, enables us to predict intent using a particle filter solely based on past motion data and tracking errors. For safe assistive control, we propose avariable impedance controller that intelligently adapts the robot’s impedance to offer assistance based on the intent estimation confidence from the DS particle filter. We validate our framework on a challenging real-world human-robot co-manipulation task and presenting promising results over baselines. Our framework represents a significant step forward in physical human-robot collaboration (pHRC), promising to extend the application of robots in various sectors by ensuring their cooperative interactions with humans are both feasible and effective.

Inspired by human-human co-manipulation, in this work, we seek to endow robots with the capability to estimate the human’s intent solely from physical guidance while considering kinematic and feasibility constraints of both the agents for fluid human-robot object co-manipulation tasks. Three main challenges need to be addressed to achieve this:

• We predict human intent, including goal and motion, in co-manipulation tasks by estimate the parameters of a 6D DS motion policy using particle filters driven by velocity tracking errors, without force-torque sensing.

• We propose a confidence-based variable impedance control scheme suited to track the estimated 6D DS motion policy. By estimating the confidence of our particle filter predictions our method fluently decides when to offer active or passive assistance without a force/torque sensor.

• We ensure goal/motion feasibility during intent estimation by trimmimg and reshaping particles using robot kinematic constraints and human manipulability measures.

• We validate our proposed constraint-aware intent estimation approach on a challenging co-manipulation task with real hardware experiments, showcasing improved performance over state-of-the-art techniques.