Smart robotic wheelchair offers enhanced autonomy and control

Recent advances in the fields of human-infrastructure interaction, electronic engineering, robotics and artificial intelligence (AI) have opened new possibilities for the development of assistive and medical technologies. These include devices that can assist individuals with both physical and cognitive disabilities, supporting them throughout their daily activities.

Researchers at the University of Michigan recently developed CoNav, a smart wheelchair controlled via a Robot Operating System (ROS) based framework. The new wheelchair, presented in a paper on the arXiv preprint server, could help to improve the quality of life of individuals who are temporarily or permanently unable to walk, allowing them to move in their surroundings more intuitively and autonomously.

“The inspiration for this work stems from a broader challenge in assistive mobility for people with disabilities (PWD),” Vineet Kamat, senior author of the paper, told Tech Xplore.

“Mobility plays a fundamental role in independent living, social participation, and overall quality of life. While powered wheelchairs (PWCs) have improved mobility for many users, they still present operational challenges, leading to difficulty in navigating complex environments.”

Most wheelchairs developed in the past are either fully autonomous or can only be manually controlled by users. Both these types of solutions have their limitations, with autonomous wheelchairs often performing poorly in dynamic environments and manually controlled ones being difficult to operate for people with mobility impairments.

The team at the University of Michigan set out to develop a new smart wheelchair that could overcome the shortcomings of existing solutions. The CoNav chair, the wheelchair they developed, allows users to partially control its operation, yet it can also navigate environments autonomously.

“Our primary objective was to design a smart wheelchair that balances autonomy and user control, enabling efficient navigation while maintaining user trust,” said Carol Menassa, senior author of the paper.

“Our approach is grounded in the idea that assistive mobility should be a partnership between the user and the intelligent system rather than a fully autonomous or fully manual process. This human-in-the-loop framework fosters greater trust, usability, and accessibility, helping PWD navigate daily life with greater ease and confidence.”

CoNav chair differs from fully autonomous wheelchairs introduced in the past, which have been found to move less effectively in complex and dynamic environments. Instead of always operating autonomously, the team’s newly developed chair enables a collaborative navigation experience.

“To realize this shared control, we adopt a modern control method named model predictive control (MPC), which acts as the intelligent decision-making system for the wheelchair,” explained Yifan Xu, lead student researcher and first author of the study. “In simple terms, by properly designing the cost function, MPC continuously predicts the best movement based on both user input and environmental data.”

The smart wheelchair developed by the University of Michigan team has various underlying components, including the chair structure, wheels, a joystick and various sensors. Users can indicate the direction in which they would like the chair to move by moving the joystick.

“The wheelchair also has sensors (LiDAR, camera, IMU) that detect obstacles, walls, and pathways,” said Xu. “Using these sensors, it generates an optimal path to a desired destination.”

Notably, the new wheelchair offers the possibility of adaptive control. Essentially, this means that instead of always following the directions provided by users via the joystick or fully ignoring them, it blends these directions into its own navigation plan.

Moreover, if users frequently provide guidance on where they would like the chair to move (i.e., actively steering the chair), the wheelchair learns to prioritize their input over the navigation plan it created. In contrast, when users stop offering directions via the joystick, the system takes over to enable safe and smooth navigation.

“CoNav has various unique advantages,” said Kamat. “Firstly, it reduces user effort, as people using it don’t need to constantly correct its movements like they would when using manual wheelchairs. In addition, it can prevent collisions, intelligently avoiding obstacles while still following the user’s intentions.”

Compared to other smart wheelchair designs introduced in the past, CoNav is designed to make users feel more in control, increasing their trust in the chair and improving their overall experience. By blending autonomous and manual control, the chair makes navigation smoother and more intuitive, reducing abrupt movements and trajectory corrections.

“By combining user intent and AI-driven navigation, the CoNav Chair creates a more natural, efficient, and user-friendly wheelchair experience, improving the mobility and independence of PWD in everyday environments,” said Menassa.

The research team evaluated their chair in a series of real-world experiments conducted in an indoor environment that included static and dynamic (i.e., moving) obstacles. Their findings were very promising, as CoNav appeared to outperform both fully autonomous and manually controlled wheelchairs in terms of efficiency, safety and user satisfaction.

“The system reduces abrupt movements, leading to a more natural and comfortable experience for users,” explained Xu. “Traditional manual wheelchairs can be difficult to control in tight spaces, leading to frequent collisions with nearby objects. In contrast, fully autonomous wheelchairs sometimes freeze when encountering obstacles in dynamic settings. Our shared control approach blends user input with real-time obstacle avoidance, leading to fewer collisions and more successful navigation.”

The researchers found that CoNav also allowed users to reach a desired destination faster than wheelchairs that can only be operated manually. The people who tried the chair during the team’s evaluations also reported feeling more comfortable and confident when using the researchers’ wheelchair, as they felt that they were still in control of its movements while also benefiting from its autonomous navigation capabilities when needed.

“The CoNav Chair could be used to support individuals with motor impairments, providing them with greater independence,” said Kamat. “In the future, it could be deployed in health care and rehabilitation centers, helping patients with temporary injuries who may not be able to use a fully manual wheelchair but still want some level of control.”

In the future, the CoNav chair could be commercialized and deployed in health care settings to assist people with temporary or permanent physical disabilities. In addition, as cities become smarter and connected, the chair could be connected to other IoT-based navigation aids, such as smart crosswalks, elevators or indoor guidance systems, allowing users to safely move in urban landscapes.

The research team is now working to further improve the capabilities of CoNav and assess its performance. They soon plan to conduct tests with a more diverse group of users, including people with varying levels of mobility impairments.

“As part of our next studies, we also plan to develop a socially aware shared control framework, allowing the wheelchair to navigate in crowded environments while considering pedestrian movement,” added Menassa. “In addition, we would like to enhance the chair’s user interface, integrating multimodal inputs (e.g., EEG signals and eye-tracking) to support a broader range of users.”

© 2025 Science X Network