PulseRide: A Robotic Wheelchair for Personalized Exertion Control with Human-in-the-Loop Reinforcement Learning (2506.05056v1)

Abstract: Maintaining an active lifestyle is vital for quality of life, yet challenging for wheelchair users. For instance, powered wheelchairs face increasing risks of obesity and deconditioning due to inactivity. Conversely, manual wheelchair users, who propel the wheelchair by pushing the wheelchair's handrims, often face upper extremity injuries from repetitive motions. These challenges underscore the need for a mobility system that promotes activity while minimizing injury risk. Maintaining optimal exertion during wheelchair use enhances health benefits and engagement, yet the variations in individual physiological responses complicate exertion optimization. To address this, we introduce PulseRide, a novel wheelchair system that provides personalized assistance based on each user's physiological responses, helping them maintain their physical exertion goals. Unlike conventional assistive systems focused on obstacle avoidance and navigation, PulseRide integrates real-time physiological data-such as heart rate and ECG-with wheelchair speed to deliver adaptive assistance. Using a human-in-the-loop reinforcement learning approach with Deep Q-Network algorithm (DQN), the system adjusts push assistance to keep users within a moderate activity range without under- or over-exertion. We conducted preliminary tests with 10 users on various terrains, including carpet and slate, to assess PulseRide's effectiveness. Our findings show that, for individual users, PulseRide maintains heart rates within the moderate activity zone as much as 71.7 percent longer than manual wheelchairs. Among all users, we observed an average reduction in muscle contractions of 41.86 percent, delaying fatigue onset and enhancing overall comfort and engagement. These results indicate that PulseRide offers a healthier, adaptive mobility solution, bridging the gap between passive and physically taxing mobility options.

• The paper introduces a novel robotic wheelchair that leverages human-in-the-loop reinforcement learning to regulate user exertion via physiological feedback.
• The system uses a Deep Q-Network and Markov Decision Process fed by ECG, heart rate, and velocity data to maintain moderate cardiovascular activity, extending active periods by up to 71.7%.
• Experimental tests indicate a 41.86% reduction in muscle strain on slate surfaces, demonstrating improved physical efficiency and adaptive assistance.

PulseRide: Adaptive Assistance Through Human-in-the-Loop Reinforcement Learning

The paper "PulseRide: A Robotic Wheelchair for Personalized Exertion Control with Human-in-the-Loop Reinforcement Learning" presents a compelling innovation in assistive technology designed to enhance mobility for wheelchair users. By employing Human-in-the-Loop Reinforcement Learning (HITL-RL), the presented PulseRide system exemplifies the integration of adaptive robotic assistance with real-time physiological feedback, notably using heart rate and electrocardiogram (ECG) data.

Summary and Technical Methods

PulseRide is configured as a robotic wheelchair that dynamically adjusts the assistance level based on users' physiological responses. Unlike traditional wheelchairs, which either provide no assistance or fixed assistance relying on user effort and manual control, PulseRide utilizes a HITL-RL approach. Specifically, it employs Deep Q-Network (DQN) reinforcement learning to optimize push assistance, keeping users within a personalized exertion range tuned to moderate cardiovascular activity. This automatic adaptation is achieved by modeling a Markov Decision Process with state inputs from encoded ECG data, heart rate, and wheelchair velocity, allowing for continuous real-time adjustments according to each user’s exertion levels.

The system includes sophisticated hardware such as DC brushed motors for propulsion, a Polar H10 heart rate sensor for physiological monitoring, and various supplementary sensors to capture wheelchair dynamics. The design ensures seamless operation, allowing users to engage in physical activity that is appropriately challenging and conducive to cardiovascular health without undue stress on musculoskeletal systems.

Key Results

The experimental results garnered from tests involving ten participants showed impressive outcomes. Notably, PulseRide maintained the users' heart rates within the moderate activity zone for up to 71.7% longer than manual wheelchairs. Furthermore, muscle contractions were reduced by an average of 41.86% on slate surfaces, showcasing a significant reduction in physical effort without compromising cardiovascular benefits. Cross-environment evaluation, conducted across varying terrains, highlights the system's adaptability by consistently maintaining moderate exertion zones despite differing surface coefficients, such as slate and carpet.

Implications and Future Directions

The implications of this paper are both practical and theoretical. Practically, PulseRide fosters a healthier lifestyle by providing a balance between the physical demands of manual propulsion and the complete reliance on powered assistance. The adaptive assistance encourages sustained engagement in physical activity, reducing risks of inactivity-induced conditions such as obesity and shoulder injuries, which are prevalent among wheelchair users. Theoretically, the application of HITL-RL into assistive robotics exemplifies innovation in how AI-driven systems can integrate human physiological feedback for task optimization.

In terms of future directions, several areas could benefit from further exploration. The paper’s findings and methods can pave the way for enhancing the PulseRide system with more advanced reinforcement learning algorithms such as Actor-Critic methods or Proximal Policy Optimization to further refine the adaptation process. Moreover, additional studies involving individuals with disabilities are essential to validate the usage scenarios, address real-world challenges, and enhance the system’s capability to effectively cater to diverse user needs.

Overall, the PulseRide system stands as a noteworthy advancement in assistive technology, providing insights and frameworks applicable to broader applications in robotics where human-centered adaptability is paramount. This paper lays critical groundwork for future research in personalized assistive systems, promising continued innovation beneficial not only to wheelchair users but potentially extending to other areas requiring adaptive robotic assistance solutions.