INTELLIGENT WHEELCHAIRS

Transcription

1 INTELLIGENT WHEELCHAIRS Patrick Carrington

2 INTELLWHEELS: MODULAR DEVELOPMENT PLATFORM FOR INTELLIGENT WHEELCHAIRS Rodrigo Braga, Marcelo Petry, Luis Reis, António Moreira

3 INTRODUCTION IntellWheels is a flexible platform for developing intelligent wheelchairs Intelligent Wheelchair is a robotic device built from an electric-powered wheelchair provided with a sensorial system, actuators, and processing capabilities

4 RELATED WORK The authors presented several Intelligent wheelchair designs and platforms for design. They indicate that the IntellWheels platform is better because it should allow: Easy Development of low-cost IWs without major modifications to existing wheelchairs Testing of new algorithms, interfaces, and allow patient training through virtual or mixed reality environments.

5 METHODS Their methods section describes the IntellWheels platform and the included modules/agents. Figure 1: Project Modules Figure 6: Hardware Framework

6 RESULTS Prototype Interface MultiModal Input Shared Control - Small User Study Planning and Autonomy Mixed Reality

7 DISCUSSION The authors explained how their results section was presented with the intent of demonstrating the platforms features and not evaluating the different components. They provide pointers to other papers where their components were evaluated and explained in more detail.

8 CONCLUSIONS Their implementation was achieved for under $4,000, including the chair and all additional hardware. They were able to describe the flexibility of their system to convert commercially available wheelchairs They described/demonstrated the use of multimodal input They described how their platform works with a mixed reality simulator.

9 INDOOR AND OUTDOOR MOBILITY FOR AN INTELLIGENT AUTONOMOUS WHEELCHAIR C.T. Lin, Craig Euler, Po-Jen Wang, and Ara Mekhtarian

10 OVERVIEW Wheelchair users can have a varying level of dependency on caregivers and other assistance. One of the goals for smart wheelchairs is to minimize or remove this dependence. This paper presents an implementation of a smart wheelchair

11 CHALLENGES BCI control allows access by users independent of their physical ability. However, sometimes there is a lag between the command and the action. Autonomous Control Eliminates gaps by continuously scanning the environment and executing actions accordingly A hybrid approach would allow the chair to respond to changes in the environment.

12 STATE OF THE ART The authors reviewed several videos of other smart BCI controlled chairs and identified several limitations to the environments tested including: Empty halls No pedestrians Empty rooms Strategically placed chairs Operating on a paved driveway Entering or exiting a door to a study room

13 STATE OF THE ART In addition the latency issue of the EEG input was not addressed. The delay can become a an issue when the user decides to turn and instead bumps into a corner of the wall or a person.

14 EXTENDING TO THE OUTDOORS This paper extends the groups previous work of the smart chair operating in a busy hallway with multiple pedestrians. Begin by Retrofitting a motorized wheelchair Continue with BCI, Autonomous, and Hybrid controls

15 CSUN SMART WHEELCHAIR A motorized wheelchair fitted with: a Laser Range Finder (LRF) a digital camera a laptop a pair of QuickSilver motors National Instrument s LabVIEW software

16 CONTROL MODES BCI Uses the Emotiv EPOC EEG headset to translate brainwaves to commands for the wheelchair Autonomous Control Senses the environment to compensate for input to action lag times Hybrid Control User input is evaluated for safety by the autonomous system before being executed. The user s input may be overridden by the computer. The user may initiate an emergency stop if necessary

17 DEVELOPMENT AND RESULTS Terrain Recognition Used learning algorithms to classify different color profiles which represented different terrains. Image Processes Used computer vision to acquire and process the necessary imagery to input to the learning algorithms Wheelchair Excursion/Travel The LRF was used to identify changes in terrain elevation and evaluate traversable areas and assist with identification of hazards and impasses.

18 CONCLUSIONS This solution can be helpful for restoring some of the independence back to the wheelchair user. Improving the processing power available to the algorithms could greatly improve performance of this solution.

19 WHY THIS RESEARCH IS IMPORTANT Intelligent wheelchairs can add another dimension of independence for people with mobility issues (Paper 2). The first paper demonstrated a platform that allows for flexibility in developing intelligent wheelchairs.

20 RELEVANCE TO OUR CLASS Situations can vary greatly from person to person making flexibility of design a very desirable feature. To put things into perspective the implementation of the IntellWheels chair cost $4,000 while the CSUN implementation cost around $10,000*. Comparing the two chairs designs one clearly stands out as different *Second phase for brainwave-operated wheelchair at CSUN-

21 QUESTIONS