1996 Project Reports


Obstacle Avoidance - training with computer simulated environments

David A. Brown, PhD; David L. Jaffe, MS; Sandy Dunn Gabrielli, BS; Douglas Schwandt, MS; and James Anderson, JEM

Objective - This pilot project pursues development of a technique to monitor, train, and improve stepping-over responses (SOR) to typical hazards encountered during walking. Targeting persons at risk of falling, the specific objective is to simulate important characteristics of obstacles encountered during walking so as to elicit SOR that are appropriate for the avoidance of tripping and/or falling. This is accomplished by measuring the SOR of young and elderly healthy subjects as they encounter real obstacles during overground walking, and comparing these SOR with those exhibited while using simulated obstacles during both treadmill and overground walking. The results are predicted to show relationships between kinematic parameters and obstacle characteristics that enable prediction of SOR patterns. In addition, we will be able to demonstrate the advantages and limitations of using treadmill walking as a substitute learning environment for overground walking.


Figure 1. Subject walking on a treadmill viewing computer simulated obstacles.

Rationale - There are many problems with the current methods of providing obstacle avoidance training to individuals at high risk for tripping or falling. Practicing these techniques with real obstacles or out-of-doors exposes the patient to the risk of injury due to a fall. Physical therapists report that they limit or don't attempt training to avoid possible injury to the patient. In addition, there is the logistical problem of providing the range of conditions that a patient might be confronted with, such as varying surface conditions (e.g. wet, bumpy, and icy) and ambient lighting conditions (e.g. dusk and darkness).

Simulated environment techniques can address some of the shortcomings of current training methods. With the proposed method, a trainee can make mistakes without fear of injury. This training can occur within a clinical setting. Earlier training may aid in recovery, and improve self-esteem and functionality. The techniques that develop from this pilot proposal will encourage repetition and practice in problem areas and the ability to control hazardous situations at a level appropriate to the individual, taking into account his/her progress. The user can get visual feedback of his/her progress. In addition, a quantitative measurement of patient progress will be obtained using a motion analysis portion of the system. Most importantly, the user will be introduced to multiple situations that induce falls so as to plan strategies for fall avoidance.

Approach - Subjects walk in two environments under one of three conditions. They walk over a 25 x 5 foot walkway platform that can be raised at one end to allow up to a 10% pitch. Portable walls, of a neutral color, are placed along the sides of the walkway. Wooden blocks have been constructed so that they can be connected to form obstacles of different heights and lengths. The obstacles are the same width as the walkway to discourage the subjects from stepping around the obstacle. They also walk on a motorized treadmill, that allows variations in pitch and speed. Three cameras mounted from the ceiling capture the bilateral sagittal-plane lower limb kinematics of subjects as they walk in the two environments (treadmill and walkway). The distances and marker placements will be calibrated and identified.

The "i-glasses" head mounted display (HMD) from Virtual I/O is used to display free-running real and computer-generated videos. Subjects are asked to fix their gaze on the ground approximately 6 to 10 feet ahead of the tester to maintain a stationary head position. Computer simulated simple rectangular obstacles of various apparent heights, lengths, and locations in the walking path have been developed on a computer using commercially available virtual reality software. The surfaces of the virtual objects are shaded, and the objects themselves appear to grow larger as they approach. This strategy provides the necessary perceptual cues for a convincing 3-dimensional image in the monocular display. Both real world objects and animations of computer simulated objects at several approach speeds and apparent tilt angles have been created and videotaped for subsequent playback through the head mounted display during testing.

Status - Currently, lab setup, display development, and the initial pilot are underway.

Future Work - If this pilot project supports the hypothesis that treadmill simulated obstacles can be used to predict real environmental obstacle responses, a Merit Review proposal will be submitted that will include subject populations the elderly and of people with gait impairments. Head tracking and real-time computer generated obstacle simulation will be pursued at that time. The major focus of the Merit Review effort will be to develop obstacle avoidance training methods with the possibility of including teaching environmental factors and modifications to avoid falls.

Future work in this project area could include employing this simulation technique with walking aids such as canes and crutches. Other potential areas of research include the study of improvements in fitness and gait through simulation of walking environments for ambulatory nursing home patients. The system could provide an enjoyable and safe "memory lane" for general exercise, a safe environment for "wanderers" or a simulated practice session for wayfinding to older patients newly admitted to nursing home facilities.

Republished from the 1996 Rehabilitation R&D Center Progress Report.