Abstract-This article reviews several tools we have developed to improve the understanding of locomotor training following spinal cord injury (SCI), with a view toward implementing locomotor training with robotic devices. We have developed (1) a small-scale robotic device that allows testing of locomotor training techniques in rodent models, (2) an instrumentation system that measures the forces and motions used by experienced human therapists as they manually assist leg movement during locomotor training, (3) a powerful, lightweight leg robot that allows investigation of motor adaptation during stepping in response to force-field perturbations, and (4) computational models for locomotor training. Results from the initial use of these tools suggest that an optimal gait-training robot will minimize disruptive sensory input, facilitate appropriate sensory input and gait mechanics, and intelligently grade and time its assistance. Currently, we are developing a pneumatic robot designed to meet these specifications as it assists leg and pelvic motion of people with SCI.

This paper presents work towards quantifying the manual assistance provided by therapists during locomotor training for people with spinal cord injury. The final goal is to translate human trainers' skill into gait-training robot algorithms. Locomotor training is a rehabilitation technique in which three therapists assist the legs and hip of the patient to walk on a treadmill while part of the patient's body weight is supported by an overhead harness. We have developed a sensorized orthosis that measures shank kinematics and therapist forces during locomotor training. The orthosis is attached to one of the legs, so that one of the therapists assists through the orthotic interface. This interface is similar to how a locomotor-training robot is attached to the patient's shank. However, the force and intelligence behind the orthosis is not robotic, but human. Our intention is to quantify and analyze the human therapists' intelligence and expertise to help design better gait-training robot control algorithms. In this paper we present some preliminary results from the first locomotor training sessions with spinal cord injured patients using this sensor system. A key initial finding is that even skilled trainers assist with substantial differences in terms of both forces and motions. With the same patient, same stepping speed and same body weight support, the differences in peak forces applied to the knee between trainers were up to 100% in some sessions.

Abstract-Whole-body shocks and vibrations experienced during manual wheelchair use can decrease an individual's comfort, increase the rate of fatigue, result in injury, and consequently limit mobility and community participation. We used a wheelchairvibration simulator to examine whether the seat reaction forces experienced by wheelchair users were differentially influenced by wheelchair suspension, trunk-muscle innervations, and ground speed. We used wheelchairs instrumented with load cells and accelerometers to determine the forces transmitted from the seat frame and the head accelerations experienced by riders. We determined that self-selected speed, seat force, and head accelerations differed between subjects with and without trunk-muscle innervations and between rigid and suspension wheelchairs. Seat force and head accelerations were greatest in the rigid-frame wheelchair and lowest in the spring-type suspension-frame wheelchairs. Those participants without trunk-muscle innervations preferred slower speeds than those with trunk-muscle innervations. Forward head accelerations were greater in those without than with trunk-muscle innervations. Wheelchair rear-suspension systems may improve wheelchair mobility function in terms of comfort at higher velocity by minimizing the seat forces and head accelerations experienced by the riders, especially those with higher level spinal cord injury and diminished postural control.

This paper overviews our recent efforts to develop robotic devices to help people relearn how to walk after spinal cord injury. Our efforts are focused on two goals. The first is to develop robotic devices that allow natural gait movements and good force control. We have developed a five degrees-of-freedom robot (PAM) that accommodates natural pelvic movement during walking. PAM uses pneumatic actuators and a nonlinear control algorithm to achieve good force control. We have also developed a novel leg robot, ARTHuR, which makes use of a linear motor to precisely apply forces to the leg during stepping. Our second goal is to develop optimal training algorithms for robotic gait training. Toward this goal, we have developed a small-scale robotic device that allows us to test locomotor training techniques in rodent models. We have also developed an instrumentation system that allows us to measure how experienced therapists manually assist limb movement. Finally, we are developing computational models of motor rehabilitation. These models suggest that assisting in stepping only as needed with a force-controlled robotic device may be an effective method for improving locomotor recovery.