Wheelchair Robot Arm: From MANUS to ABB Cobot Safety
How the wheelchair robot arm grew from MANUS and Kinova JACO to AI wheelchairs in 2026, and what ABB GoFa force limiting and hand guiding could add.
INDUSTRIAL ROBOTICS
Chat With Robot
10/1/20266 min read
For someone who drives a power wheelchair but has little strength in their arms, the hardest part of the day is often reaching: a cup on a high shelf, a door handle, a phone on the floor. A wheelchair robot arm is mounted on the side of the chair and steered with the same joystick, so the user can pick things up, drink and open doors without waiting for help. This post looks at how these arms began with MANUS in the Netherlands and Kinova's JACO in Canada, what is being built and tested in 2026, and which features of ABB's collaborative robots, such as power and force limiting and hand guiding, matter for an arm that works a few centimeters from a person.


A NIST research assistant shows the prototype HLPR robotic chair lifting its user to reach a high shelf, 2011. Photo: National Institute of Standards and Technology via Wikimedia Commons (public domain)
How MANUS and JACO put an arm on the chair
Before any of this, help with reaching came from people. A wheelchair user with weak hands needed a relative, an attendant or a nurse for meals, drinks, doors and dressing, and many still do. Simple tools filled some gaps, such as long reachers with a trigger and jaws, mouth sticks and trays fixed to the chair. Powered wheelchairs brought independence in moving around. In the early 1950s the Canadian engineer George Klein built one of the first electric wheelchairs at the National Research Council of Canada, meant for people paralyzed from the neck down, many of them war veterans, though a motor in the chair did nothing for reaching and grasping.


Robert Lee Donley, a miner injured at the Amonate Mine in Virginia, in his wheelchair beside a bed, 1946. Photo: Russell Lee / US National Archives via Wikimedia Commons (public domain)
The first wheelchair arm made in real numbers came from the Netherlands. MANUS, later sold as the Assistive Robotic Manipulator by Exact Dynamics in Didam, was developed from the mid 1980s and went into production in the early 1990s. It folded away beside the chair and was driven with a keypad or joystick, one joint or one direction at a time. Dutch users could get one through the public health insurance system after a prescription and assessment process. Exact Dynamics later followed MANUS with a newer model, the iARM.


An iARM assistive robot arm, the successor to MANUS, mounted on a power wheelchair. Photo: Assistiveinnovations / Wikimedia Commons (CC BY-SA 3.0)
Canada came next. Charles Deguire and Louis-Joseph L'Écuyer founded Kinova near Montreal in 2006. Deguire's uncle had built his own manipulator for his wheelchair, and that idea became the company's starting point. By about 2010 Kinova was selling JACO, a light six axis arm with a three finger hand that mounts on power wheelchairs. Users drive it with the chair's joystick and switch modes to move the hand, turn the wrist or close the fingers. JACO arms also turn up in many university labs that study assistive robots.


Admiral Mike Mullen tries a robotic wheelchair prototype at a wheelchair research center in Pittsburgh, April 2010. Photo: U.S. Department of Defense via Wikimedia Commons (public domain)
Wheelchair arms and home robots in 2026
Research now aims at autonomy. Driving a six axis arm joint by joint with a joystick is slow and tiring, so labs add cameras and AI that let the user pick an object and leave the motion to the robot. In a Center for Data Innovation interview published on 10 September 2026, Owen Kent, CEO of the Colorado company ATDev and a lifelong wheelchair user with muscular dystrophy, described a robotic wheelchair with sensors and an integrated robotic arm for daily tasks at home. ATDev is developing it as the Robotic Assistive Manipulation and Mobility Platform, with a grant from ARPA-H, the US Advanced Research Projects Agency for Health.


A Kompai assistive robot, built to help frail people, with older residents. Photo: Kompai / Wikimedia Commons (CC BY-SA 4.0)
A second approach leaves the chair alone. On 28 September 2026 IEEE Spectrum profiled Charlie Kemp, cofounder and chief technology officer of Hello Robot, whose Stretch is a light mobile manipulator with its own wheeled base and a telescoping arm. Kemp describes how working with people with disabilities shaped the way the company designs the robot. The two designs trade off differently. A separate robot can fetch things from another room while the user stays put, and an arm on the chair goes wherever the user goes, outdoors and into a van included, but adds weight and width to the wheelchair.


NIST research assistants demonstrate the HLPR robotic chair prototype helping with leg exercise, 2011. Photo: National Institute of Standards and Technology via Wikimedia Commons (public domain)
ABB does not make a wheelchair arm, and its cobots are far too heavy to hang on a chair: a GoFa CRB 15000 weighs 28 kg. Its safety design still deals with the problems an assistive arm faces. GoFa has torque sensors in all six joints, so it can limit the power and force it applies and stop when it touches a person. It can be pushed by hand into a new pose, which ABB calls lead through, and the Wizard tool lets people without programming experience build a task from graphical blocks, as covered in our guide to no code robot control.


Moving a GoFa cobot arm by hand. Torque sensors in its joints let it feel the push. Photo: ABB
What cobot safety could bring to the next wheelchair arms
Cost is the first barrier at home. Assistive arms are made in small numbers, and whether a public insurer, a charity or the user pays depends on the country and often on a long application. Factory cobots are sold in far larger volumes, and that market has made torque sensors, compact motors and safety controllers cheaper and better tested. Some of those parts and methods could reach wheelchair arms. How far that would lower the price of an assistive arm is still an open question, and the user base stays small.


Guiding a GoFa cobot by hand over a work table, seen from above. Photo: ABB
Safety is the second barrier. Factory cobot rules assume that a worker can step back from the robot, while a wheelchair user with weak arms may not be able to move away at all, and the arm works close to the face. Assistive arms are usually regulated as medical devices, and home service robots more generally fall under the personal care robot standard ISO 13482. Several cobot features could carry over: speed and force limits per zone, like those set with ABB SafeMove on industrial robots, collision detection from joint torque, and hand guiding so a carer can push the arm out of the way. Occupational therapists also need training to fit and adjust these arms.


Building a cobot program from Wizard blocks on the FlexPendant. Photo: ABB
Factories are part of this story too. A cobot that is safe to touch can hand parts to a seated worker, hold a workpiece at a comfortable height or take over the lifting in a task, which opens more jobs to wheelchair users. Workstations like these are usually laid out and checked in simulation first, for example in RobotStudio, where reach, height and speed limits can be tried before anyone sits next to the arm. Wheelchair arm designers have worked on the same problem since MANUS: fitting an arm to one person's reach and strength.


Trying on an Ekso Bionics upper body exoskeleton in 2020. Photo: Office of Rep. Mark DeSaulnier / Wikimedia Commons (public domain)
Innovation
AI solutions for effortless ABB robot control.
Automation
Robotics
ceojohntran@chatwithrobot.net
+84905311611
© 2025. All rights reserved.
qtran1215@gmail.com