PulseRide: A Robotic Wheelchair for Personalized Exertion Control with Human-in-the-Loop Reinforcement Learning Volume: 0 0 0 0 Conference: Make email; Jun 24 2025 - Jun 26 2025; Manhattan, New York CCS: Human-centered computing Accessibility technologies
Azizul Zahid, Bibek Poudel, Danny Scott, Jason Scott, Scott Crouter, Weizi Li, Sai Swaminathan · 2025
Read PulseRide: A Robotic Wheelchair for Personalized Exertion Control with Human-in-the-Loop Reinforcement Learning Volume: 0 0 0 0 Conference: Make email; Jun 24 2025 - Jun 26 2025; Manhattan, New York CCS: Human-centered computing Accessibility technologies by Azizul Zahid, Bibek Poudel, Danny Scott, Jason Scott, Scott Crouter, Weizi Li, Sai Swaminathan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-05056.txt.
Introduction
Azizul Zahid and co-authors present PulseRide, a wheelchair that meters out motor help to keep its user in a healthy exertion zone. Power chairs invite inactivity and weight gain. Manual chairs wreck shoulders over time. PulseRide reads heart signals and chair speed, then tunes its push assistance moment by moment.
A reinforcement learning system learns each rider's responses. In tests with ten users over carpet and slate, it held heart rates in the moderate zone up to 71.7 percent longer than manual chairs. Muscle contractions fell by about 42 percent on average. The authors frame the chair as a middle path between passive rides and punishing ones.
Catalog record
| CREATOR | Azizul Zahid, Bibek Poudel, Danny Scott, Jason Scott, Scott Crouter, Weizi Li, Sai Swaminathan |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 55 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-05056
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
PulseRide: A Robotic Wheelchair for Personalized Exertion Control with Human-in-the-Loop Reinforcement Learning Volume: 0 0 0 0 Conference: Make email; Jun 24 2025 - Jun 26 2025; Manhattan, New York CCS: Human-centered computing Accessibility technologies Azizul Zahid, Bibek Poudel, Danny Scott, Jason Scott, Scott Crouter, Weizi Li, Sai Swaminathan 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, with shoulder pain affecting 42-66%. 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% longer than manual wheelchairs. Among all users, we observed an average reduction in muscle contractions of 41.86%, 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. 1. Introduction Figure 1. Comparison of wheelchair assistance paradigms. Manual wheelchair (top) provides no assistance and requires maximum user effort, leading to early onset of fatigue and risk of over-exertion. Power wheelchair (bottom) offers fixed assistance but discourages physical activity i.e., complete reliance on motorized propulsion. PulseRide (middle) provides personalized adaptive assistance, balancing user’s physical effort (to maintain moderate cardiovascular exertion) and mobility needs. For the same effort budget, as PulseRide delays the onset of fatigue, users are able to cover more distance (d2>d1). Global demand for assistive technologies is substantial and growing. The World Health Organization (WHO) estimates that 2.5 billion people need one or more assistive products, a figure projected to reach 3.5 billion by 2050 ( Organization and Fund 2022 ) . A significant portion of this need is related to mobility impairment, requiring assistive devices like canes, walkers, and wheelchairs ( Sehgal et al. 2021 ) . In particular, 80 million people rely on wheelchairs for daily mobility ( World Health Organization 2023 ) , essential for both routine tasks (e.g., moving around the home) and broader social participation. While standard manual wheelchairs provide essential mobility assistance, they rely on user effort for propulsion. Carpets, rough terrain, and even small inclines or slopes increase the effort and energy cost of mobility ( Holloway et al. 2015 ) . A growing body of evidence suggests that the repetitive stress of manual propulsion has serious health consequences. The continuous repetitive motion often leads to upper extremity injuries, with shoulder pain reported in 42% ( Dalyan et al. 1999 ) to 66% of users ( Fullerton et al. 2003 ) and common injuries to the rotator cuff muscles ( Akbar et al. 2010 ) . Additionally, bilateral carpal tunnel syndrome is frequently associated with handrim manipulation and engagement ( Asheghan et al. 2016 ) . These injuries could be mitigated by reducing push frequency to decrease the repetitive strain associated with long-term wheelchair use ( Boninger et al. 2005 ) . Power wheelchairs offer an alternative by providing motorized assistance, alleviating the physical strain of manual propulsion. However, they often provide fixed levels of support focused solely on mobility, addressing the basic requirement of independent movement from point A to B, but generally overlook the user’s need for physical activity and its cardiovascular benefits ( Edwards and McCluskey 2010 ; Andrabi et al. 2022 ) . Physical inactivity disproportionately affects wheelchair users, contributing to obesity, deconditioning, and further health decline ( Cooper et al. 1999 ) . Additionally, power wheelchair users may experience limitations in community participation ( Giesbrecht et al. 2009 ) , as reduced physical activity can hinder engagement in social settings, further impacting their well-being. A third option has emerged to address the limitations of conventional manual and power mobility: power assist wheels. For instance, systems such as SmartDrive, Yomper ( Acekare 2023 ) and e-motion ( Alber USA 2023 ) have shown to allow users to stroke the wheelchair pushrims to activate small, lightweight motors that drive the wheels for a brief period of time (seconds). Users must continue to stroke the pushrim as they would if they were propelling a conventional manual chair. This method requires less strength and endurance than manual wheeling and is especially helpful on inclines, uneven terrain, and carpeted surfaces. However, most power-assist systems still operate with a “one-size-fits-all” approach to assistance, lacking automatic adaptation to varying environmental conditions (e.g., slopes, ramps, gravel) and overlooking the variability in individual user capabilities and contexts. Assistance levels are typically adjusted manually using a knob with preset velocity settings, making these systems less responsive to real-time user and environment needs. These limitations highlight a critical gap in assistive technology that could dynamically personalize assistance to match each user’s unique ability ( Wobbrock et al. 2011 ) , physiological state and environmental context. Additionally, wheelchair users often have limited opportunities for physical activity, with manual propulsion remaining one of the few accessible forms of exercise ( Coulter et al. 2011 ) . Integrating physical activity into daily routines, beyond structured leisure time, is increasingly recommended to promote holistic health benefits ( Cowan et al. 2022 ) . Therefore, determining the appropriate level and timing of automatic assistance to meet both mobility and health goals remains an open question. Additionally, understanding how an autonomous system could achieve physical activity goals as part of daily mobility for wheelchair users is still underexplored. In response, we propose a new assistance paradigm for wheelchair users that integrates real-time physiological feedback with autonomous control to deliver personalized, exertion-aware assistance. As shown in Figure 1 , this approach provides a balanced solution between manual wheelchairs that require maximum user effort and power wheelchairs that discourage physical activity. Our research focuses on the following questions: • RQ1 : How can we design an autonomous wheelchair system that promotes physical activity while supporting users’ daily mobility? • RQ2 : How can we personalize assistance levels by dynamically adapting to users’ physiological states? • RQ3 : How can we ensure that this adaptive assistance is effective across diverse environments? To address these research questions, we designed a novel, autonomous wheelchair system called PulseRide. PulseRide utilizes Human-in-the-Loop Reinforcement Learning (HITL-RL), integrating real-time data on users’ physiological state including heart rate and electrocardiogram (ECG) signals, and wheelchair movement (speed) to dynamically adjust the timing and duration