 |
|
|
|
|
|
Pedestrian mobility is one of the most basic requirements of an independent life style. Reduction in mobility due to physical frailty or psychological factors (such as fear of falling) leads to decreased quality of life and deterioration of the body. We are developing a mobility aid, based on a wheeled walker, to augment the capabilities of its user. That is, it will allow the elderly to use the skills they possess, while supplementing those that may have declined due to age or fatigue. The key to the design of our mobility aid is that it gives control to the user, acting as any other walker, while the user is not in a difficult or dangerous situation. When the situation changes, the walker’s control system can begin to take some of the burden of steering off the user. The fundamental goal of this work is to develop a system of shared control for personal mobility aids that helps the user without making them feel as if they are not in control or being “lead about”.
Our intelligent walker can assist its user in a number of ways. First, it can provide basic safety services, such as obstacle avoidance and drop-off detection. Second, the walker can assist the user in situations that are traditionally difficult to navigate with a walker, such as tight spaces. By making minor corrections in the user’s steering input, the walker can facilitate passing through doorways and reduce the chance of catching a wheel on an object. Finally, the walker can assist the person with following a particular route through an environment. Exactly which capabilities the walker exhibits at any time depend on the will and abilities of the user.
As the demographics of the United States, and most industrialized countries, continue to shift toward an increasing contingent of elders, it is imperative that there are ways for these people to maintain their independence for as long as possible. In data obtained during the 2000 census, people aged 65 and over, comprised 12.4% of the US population in 2000. This translates to about 1 in every 8 Americans qualifying as a senior citizen. This number will continue to rise as the “baby boom” generation continues to age, and in 20 years the percentage of elders in the American population will be 25%. Thus, now more than ever, there is a definite need to examine the quality of life of the elders as they face more challenges to maintain their independence. One major challenge for the elders to overcome has been vision impairment. Three out of every four blind people are senior citizens. In addition, diseases and conditions such as macular degeneration or cataracts can cause serious vision impairment in the elders. When this factor is placed in context with hearing and memory loss as well as loss of motor coordination, doing everyday tasks and mobility become serious challenges.
|
|
|
|
|
There are many different types of aids that can help the elders maintain some of their independence and mobility. Some examples include Rollator walkers, motorized scooters, canes, walking sticks and even Seeing Eye dogs. However, these items are all limited in the amount of help that can be provided to the person. The canes, walking sticks and Seeing Eye dogs do not provide stable platforms to prevent falling as the result of a trip. Rollators and motorized scooters, though they provide stable platforms, do not address the issues of vision impairment or, to some extent, motor coordination losses.
The Medical Automation Research Center (MARC) at the University of Virginia has developed a prototype walker that is geared toward meeting those needs in a way that would help the elders and others suffering from vision impairment maintain their independence and mobility.
The current Walker prototype, which is built using off the shelf components and a three-wheel Rollator walker frame, employs sensor technology to detect obstacles, as well as a passive shared- control navigation system that takes control of the steering wheel only when necessary to avoid collision with obstacles. This, in addition to the fact that there is no drive capability, makes the walker’s users feel that they are in control rather than being led by, or having to chase after, the walker. The walker has also been augmented with rear-wheel collision detection sensors that warn the user when the back wheel may get caught by an obstacle; a problem associated with three- wheel walkers that have a wide wheelbase, particularly when negotiating doorways.
|
|
|
|
|
MARC Robotic Walker has proven useful for elders in avoiding obstacle during user-trial. Naturally, the reduction in collisions with obstacle on the course was most statistically significant, after turning on the obstacle-avoidance navigational aid, for elders with vision problems.
The prototype is not collapsible, since it incorporates off-the-shelf frame, however, the final production model includes plans to shrink the existing technology, decrease the weight, incorporate a folding seat for the user to rest and a add system of hinged joints to provide the collapsibility and portability required.
Moreover, user trials have revealed the need for a walker with higher load capacity, and more sophisticated navigation and drive system for different applications and user groups. In another version of the Robotic Walker, we intend to incorporate means to infer User Intent, using sensor technology, an optional obstacle detection module for the visually impaired, and a more involved fuzzy navigation system that will control the walker platform in steering, power assisted forward – backward drive and active braking. The navigation system will automatically vary the level of intervention according to the context inferred from both the user intent and the obstacle layout in the environment; thus implementing a truly intelligent shared control system. Such a load carrying mobility aid will be useful for elders who would want to exercise outdoors (e.g. play golf, longer walks outdoors) and/or carry heavy loads (e.g. golf clubs, shopping bags). The addition of the sensory obstacle detection module permits the visually challenged users to utilize the platform for shopping and load delivery purposes. The active braking will stabilize the platform on slopes and will provide better support for users, while a folding seat will allow them to rest when they feel tired.
Passive robots are robots that can steer their joints, but require a human to move them. The walker is a passive robot because it can only control the orientation of its front wheel and cannot move forward on its own. Little work has been done in applying passive robotics to the field of assistive technology. We feel that such devices offer a decisive advantage to the elderly because they leave final control in the hands of the user. We are currently studying how the walker can steer to achieve the movement goals of the user without compromising the balance and support that the user expects the walker to provide.
The walker’s control system is essentially an intelligent agent whose goals are to provide support and navigation assistance to its user. The agent watches the movement of the walker frame (which since it is a passive robot comes from the human) and the sensor systems in order to derive a model of the user and where they may be going. When the control agent is actively helping the user steer, there are two agents (the control system and the human) that are inputting control signals to the walker frame. Each agent learns something about the other by observing the resulting motion that the frame makes. The more accurately the control agent can predict the user’s goals, the more seamlessly it can assist in reaching them. Having the control agent formulate the user’s goals without explicit commands from the user means that the user can operate the walker in the normal way (they just push it!). There is no additional cognitive load when operating the walker and no special training is required.
The walker control agent’s goal is ultimately to help the user. There for the control agent must be submissive, but alert. When the control agent incorrectly determines the user’s goal location and the user and the control agent express different desires for how to steer the walker frame, the control agent must relent (unless the user’s command would cause them harm). When the user is in no danger, or when the control agent’s steering commands have been resisted, the agent remains quiescent. However, it remains alert, gathering sensor data to update its model of the world and the user’s goals.
The degree of control exhibited by the walker control agent depends on the abilities of the user at the time. As the user’s capabilities vary from day-to-day or hour-to-hour, the walker can take a more or less active role in the guidance of its user. Deciding where and when to change the degree of autonomy given the control agent is an active area of research. Currently, the walker gives full control to the user when no danger or difficulty is detected in the environment. Control becomes a blend of user and walker control agent commands when the walker attempts to steer, either around objects or through difficult spaces. The walker control agent takes full control when a collision or drop-off is imminent and the brakes must be applied.
In order to conduct clinical trials of our ideas on shared control, we have developed a prototype walker based on a conventional 3 wheeled frame from InvaCare. We have augmented this frame with sensors and actuators to control the front wheel and the brakes. The specific systems are discussed below.
Currently, we use a laser scanner from the SICK corporation mounted on the front of the walker. This gives us a depth map for a 180 deg. field-of-view in front of the walker. IR sensors from Sharp will be installed to monitor for collisions with the back wheels. We also have encoders on all three wheels. This allows us to determine what path the walker has been following and where it will go next.
The drive system is a belt driven stepper motor system that can turn the front wheel to a given heading. Since the drive system can only effect the orientation of the front wheel, the user must provide the motive force. In other words, the walker is a passive robot and cannot move on its own; it can only steer. Having the walker move at the user’s pace allows them to feel in control.
Currently, the walker has hand-brakes. A motorized brake system has been designed to activate the existing brakes on the rear wheels. This allows the walker to stop in dangerous situations while allowing the user to stop the walker in the conventional way.
The walker is also being outfitted with handle sensors to detect the force places on each handle. This is an indication of the movement desired by the user, but it can be detected before the walker frame actually moves.
Stability Margin Monitoring in Steering-Controlled Intelligent Walkers for the Elderly
Majd Alwan, Prabhu Jude Rajendran, Alexandre Ledoux, Cunjun Huang, Glenn Wasson, Pradip Sheth.
AAAI Fall 2005 Symposium (EMBC). Nov. 2005.
Shared Navigational Control and User Intent Detection in an Intelligent Walker
Cunjun Huang, Glenn Wasson, Majd Alwan, Pradip Sheth, Alexandre Ledoux
AAAI Fall 2005 Symposium (EMBC). Sep. 2005.
Characterization of Infrared Range-Finder PBS-03JN for 2-D Mapping
Majd Alwan, Matthew B Wagner, Glann Wasson, Pradip Sheth.
Proceedings of 2005 IEEE International Conference on Robotics and Automation (ICRA '05), Barcelona, Spain Apr. 18-22, 2005.
Passive Derivation of Basic Walker-Assisted Gait Characteristics from Measured Forces and Moments
Majd Alwan, Glenn Wasson, Pradip Sheth, Alexandre Ledoux, Cunjun Huang. IEEE Transactions on Engineering in Medicine and Biology (EMBC). Sep. 2004.
A Physics-Based Model for Predicting User Intent
in Shared-Control Pedestrian Mobility Aids
Glenn Wasson, Pradip Sheth, Cunjun Huang, Alexandre Ledoux, Majd Alwan.
Proceedings of 2004 IEEE/RSJ International Conference on
Intelligent Robots and Systems
September 28 - October 2, 2004, Sendai, Japan
User Intent in a Shared Control Framework for Pedestrian Mobility Aids
Glenn Wasson, Pradip Sheth, Majd Alwan, Kevin Granata, Alexandre Ledoux, Cunjun Huang.
Proceedings of the 2003 IEEE/RSJ International Conference on Robots and Systems, Las Vegas, October 2003.
Effective Shared Control in Cooperative Mobility Aids
G Wasson, J Gunderson, S Graves and Robin Felder. 2001. FLAIRS ’01: 509-513.
An Assistive Robotic Agent for Pedestrian Mobility
G Wasson, J Gunderson, S Graves and Robin Felder 2001.
International Conference on Autonomous Agents. In press.
Click here to see a recent news story about the walker from NBC in Charlottesville
