Che Fai Yeong

ReachMAN: a personal robot to train reaching and manipulation

Robotic devices able to train both reaching and manipulation are often large and complex and thus not suitable for decentralized use at home or in local rehabilitation centers. This paper describes a compact device with only three degrees of freedom (DOF) to train reaching and manipulation critical to activities of daily living. The design considers only the DOF necessary to train tasks such as pick-and-place of objects, drinking, eating and knob manipulation, based on low-dimensional synergies used in these tasks. Specifications from measured biomechanical parameters yield safety and suitable performance. A prototype demonstrates some of the resulting functions and therapeutic possibilities offered by this design.

Analysis of pick-and-place, eating and drinking movements for the workspace definition of simple robotic devices

Current robotic devices for rehabilitation after stroke are often large and complex as they are conceived to train arbitrary movements in the 3D space. This paper analyzes the minimal requirements that the workspace of a robotic device should have in order to promote training of principal activities of daily living, considering the shoulder movements limitations of subacute patients. Pick-and-place, drinking and eating movements of five healthy subjects were analyzed. In all these three tasks, approximately 82% of all trials deviate laterally less than 5% of the target distance and we recommend target distances of less than 40% of the arm length in order to minimize improper shoulder movements. The study can be applied for designing simpler, yet efficient robotic devices for rehabilitation of the upper limb or for constraining exercises on existing ones when dealing with stroke patients, especially for those prone to shoulder complications (e.g. subacute patients).

Effect of Grip Force and Training in Unstable Dynamics on Micromanipulation Accuracy

This paper investigates whether haptic error amplification using unstable dynamics can be used to train accuracy in micromanipulation. A preliminary experiment first examines the possible confounds of visual magnification and grip force. Results show that micromanipulation precision is not affected by grip force in both naive and experienced subjects. On the other hand, precision is increased by visual magnification of up to 10×, but not further for larger magnifications. The main experiment required subjects to perform small-range point-to-point movements in 3D space in an unstable environment which amplified position errors to the straight line between start and end point. After having trained in this environment, subjects performing in the free conditions show an increase in success rate and a decrease in error and its standard deviation relative to the control subjects. This suggests that this technique can improve accuracy and reliability of movements during micromanipulation.

ReachMAN to help sub-acute patients training reaching and manipulation

This paper presents the control strategy and efficient tasks for training with ReachMAN, a compact, portable 3 degree-of-freedom robot to train reaching, pronosupination and grasping, independently or in combination. A pilot study was performed with three sub-acute patients to evaluate the potential use of ReachMAN as a rehabilitation tool, and determine how it should be used. All subjects improved their motor function, and gains in the range and quality of movements were seen, which are not detectable by typical functional assessment.

Development of CR2-Haptic: A compact and portable rehabilitation robot for wrist and forearm training

Stroke has now become the leading cause of severe disability. Rehabilitation robots are gradually becoming popular for stroke rehabilitation to improve motor recovery, as robotic technology can assist, enhance, and further quantify rehabilitation training for stroke patients. However, most of the available rehabilitation robots are complex and involve multiple degrees-of-freedom (DOFs) causing it to be very expensive and huge in size. Rehabilitation robots should be useful but also need to be affordable and portable enabling more patients to afford and train independently at home. This paper presents a development of an affordable, portable and compact rehabilitation robot that implements different rehabilitation strategies for stroke patient to train forearm and wrist movement in an enhanced virtual reality environment with haptic feedback.

A novel haptic interface and control algorithm for robotic rehabilitation of stoke patients

Rehabilitation robots are gradually becoming popular for stroke rehabilitation to improve motor recovery. By using a robot, the patient may perform the training more frequently on their own, but they must be motivated to do so. Therefore, this project develops a set of rehabilitation training programs with different haptic modalities on Compact Rehabilitation Robot (CR2) - a robot used to train upper and lower limbs reaching movement. The paper present the developed haptic interface, Haptic Sense with five configurable haptic modalities that include sensations of weight, wall, spring, sponge and visual amplification. A combination of several haptic modalities was implemented into virtual reality games, Water Drop - a progressive training game with up to nine levels of difficulties that requires user to move the cup to collect the water drops.

Portable and Reconfigurable Wrist Robot Improves Hand Function for Post-Stroke Subjects

Rehabilitation robots have become increasingly popular for stroke rehabilitation. However, the high cost of robots hampers their implementation on a large scale. This paper implements the concept of a modular and reconfigurable robot, reducing its cost and size by adopting different therapeutic end effectors for different training movements using a single robot. The challenge is to increase the robot’s portability and identify appropriate kinds of modular tools and configurations. Because literature on the effectiveness of this kind of rehabilitation robot is still scarce, this paper presents the design of a portable and reconfigurable rehabilitation robot and describes its use with a group of post-stroke patients for wrist and forearm training. Seven stroke subjects received training using a reconfigurable robot for 30 sessions, lasting 30 min per session. Post-training, statistical analysis showed significant improvement of 3.29 points (16.20%, p = 0.027) on the Fugl-Meyer assessment scale for forearm and wrist components. Significant improvement of active range of motion was detected in both pronation-supination (75.59%, p = 0.018) and wrist flexion-extension (56.12%, p = 0.018) after the training. These preliminary results demonstrate that the developed reconfigurable robot could improve subjects’ wrist and forearm movement.

Effect of Intermittent Haptic Disturbance in Motor Skill Training

The aim of this study is to evaluate the effectiveness of haptic disturbance where the intermittent force field will applied in training motor skill. A test group (4 subjects) was trained using intermittent divergent force field while a control group (4 subjects) was trained in null force field. The parameter such as error trajectories, motion smoothness and the trial time were measured and a statistical analysis was done in order to identify motion improvement when using the proposed training method. Both groups showed motion improvement in reducing mean error, mean number of zero crossing (indicate the motion smoothness) and mean trial time after training. Only the motion smoothness in the test group improved significantly after training, in contrast to the control group, which showed no significant difference.

Accurate micromanipulation induced by performing in unstable dynamics

This study examines effects of learning 3D micromanipulation in an unstable dynamic environment. A test group trained in an unstable divergent force field while a control group trained the movement in the null force field. The subjects in the test group increased the success rate, in contrast to the control group which had similar rate after training. The error and its standard deviation decreased in the test group but not in the control group. In summary, training in unstable dynamics enable subjects to become more accurate, in contrast to training using only visual feedback.

Gaussian pedestrian proxemics model with social force for service robot navigation in dynamic environment

Pedestrian motion behaves stochastically, causing difficulties in modelling the appropriate proxemics for effective and efficient service robot navigation. Intruding the pedestrian social space can affect the social acceptance of a service robot. In this paper, a new proxemics model, Social-Force Gaussian Pedestrian Proxemics Model is presented to model the pedestrian social space and to improve the service robot navigation in dynamic human environment. This model was simulated and validated in a pedestrian simulator with both low and high pedestrian density environments. Results showed that the proposed model (i) improved proxemics representation of pedestrians, (ii) enhanced the robot performance in respecting the social norm and (iii) increased the efficiency in achieving a given task. This paper also presents the methods for parameter selections for the model without the requirement of complex tuning.