Eiichi Ohki

An algorithm of walk phase estimation with only treadmill motor current

To develop a gait rehabilitation robot for hemiplegic patients, quantitative evaluations of patient ability is needed. Patients walk phase, which includes time balance of stance and swing legs, is one of the most useful indexes. However, conventional methods measuring walk phase require laborious preparations. In this paper, a novel algorithm estimating walk phase on a treadmill by observing DC motor current is proposed. In comparison of this algorithm and conventional methods, it was verified that the proposed algorithm had as the same accuracy as foot switch. Moreover, the proposed algorithm could estimate stance phase in 0.2 (s) errors between measurements of force plate mostly (4 out of 5 healthy subjects). However, result from the 5th subject showed that the proposed algorithm had erroneously identified stance phase as swing phase when little body weight loaded on leg. This characteristic is often observed in hemiplegic gait. Therefore, the proposed algorithm might need improvement of motor current threshold. However, this algorithm had capable of estimating the time of loading body weight on leg, and thus could be useful as a quantitative evaluation tool.

Leg-dependent force control for body weight support by gait cycle estimation from pelvic movement

A movable body weight support (BWS) system for overground walking has limitations with respect to power and space for actuators. To solve this problem, a novel method by which to estimate the gait cycle from pelvic movement and feed forward control for leg-dependent force control are proposed. Based on an experiment on gait cycle estimation, a method of estimating heel contact timing from pelvic rotation is proposed. In addition, based on an experiment on leg-dependent force control, the developed feed forward control method is evaluated based on the vertical floor reaction force.

Treadmill motor current value based walk phase estimation

We have developed a gait rehabilitation robot for hemiplegic patients using the treadmill. A walk phase, which includes time balance of stance and swing legs, is one of the most basic indexes to evaluate patients’ gait. In addition, the walking phase is one of the indexes to control our robotic rehabilitation system. However, conventional methods to measure the walk phase require another system such as the foot switch and force plate. In this paper, an original algorithm to estimate the walk phase of a person on a treadmill using only the current value of DC motor to control the treadmill velocity is proposed. This algorithm was verified by experiments on five healthy subjects, and the walk phase of four subjects could be estimated in 0.2 (s) errors. However, the algorithm had erroneously identified a period of time in the stance phase as swing phase time when little body weight loaded on the subject’s leg. Because a period of time with little body weight to affected leg is often observed in a hemiplegic walk, the proposed algorithm might fail to properly estimate the walk phase of hemiplegic patients. However, this algorithm could be used to estimate the time when body weight is loaded on patient legs, and thus could be used as a new quantitative evaluation index.

Fundamental study of force control method for pelvis-supporting body weight support system

An active body weight support (BWS) system, which unloads body weight with pelvic support, has been developed to assist the walking movement. This system unloads body weight with a motor-actuated device from below with pelvic support, unlike prevailing BWS systems that lift up the subject from above via a harness connected to a wire. The force control method to unload body weight has not been sufficiently studied. As a first step, a comparison study between normal walking and walking with the developed BWS mechanism was conducted to specify the force control method. Since the precise and constant unloading force is believed to be an important prerequisite for BWS gait therapy, different constant unloading forces were set as the target unloading force. The target unloading force was varied from 100 (N) to 300 (N) to observe the difference among forces. The measured unloading forces were not completely constant but fluctuated. The motor introduced a delay at over 200 (N) during lifting. As a result, the floor reaction force was reduced by the target unloading force. However, some differences were found in the bimodal shape of the floor reaction force and the trajectory of the sacrums vertical position. The results showed the necessity of damping the fluctuation of unloading force by improving following characteristics and adjusting target velocity. Further precise force control will be carried out to realize constant unloading force.

Treadmill motor current based real-time estimation of anteroposterior force during gait

We have been developing a new vehicle, “Tread-Walk 2”, which supports walking for elderly. A control algorithm to improve the operability has been constructed. As the first step, we accurately estimated the users anteroposterior force from the motor current value without the force sensor. This method is to develop a new mechanical model that considers the friction forces of the treadmill and remove the modeled friction losses from the output. However, we need the vertical force in order to develop a mechanical model. Thus, we proposed the new method to estimate the vertical force without the force sensor. This paper describes the new method to approximate the waveforms of the vertical forces as square waves by adding the users weight as a parameter to the stance phase. By comparing the estimated anteroposterior force using the new method with the measured one using the force plate, the waveform pattern of the estimated one was similar with that of the measured one in two young subjects whose physical characteristics were different. This showed that the proposed method might possibly be useful for estimating the anteroposterior force in real-time.

A Gait Phase Measurement System Using Treadmill Motor Current

Abstract The article describes the development of a gait phase time-based split-belt treadmill measurement system. Conventional methods of measuring gait phase, such as the foot switch and force plate, require significant preparation and are costly. In this article, we propose a simple, cheap, and accurate gait phase measurement system that utilizes only the treadmill motor current value. Comparison of this algorithm with conventional methods reveals that the proposed algorithm is as accurate as the foot switch. Moreover, the proposed algorithm can estimate stance phase within a 0.2 s error of the measured value of the force plate in most cases (four out of five healthy subjects). This accuracy is higher than that of the foot switch which is widely used in the clinical field.

Split belt treadmill with differential velocity and biofeedback for well-balanced gait of patient with stroke

A split belt treadmill robot for gait rehabilitation was developed to improve the symmetry of the stance phase time of patients with stroke. The system, which increases the stance phase time of the affected leg and then realizes a well-balanced gait, is divided into two components. First, the stance phase of the unaffected (“sound”) and affected legs is measured and presented visually in real time to the patient and physical therapist as biofeedback. Second, using the biofeedback of the stance phase, the physical therapist sets two different velocities of the treadmill belts for the sound and affected legs. In an experiment, eleven patients with chronic stroke participated in a short-term intervention trial (twenty gait cycles) of the developed treadmill system. Three of the five subjects who had lost balance between the stance phase of the sound leg and that of the affected one improved their gait balance in the intervention trial. In addition, one subject kept the well-balanced gait after the intervention. In the future, the algorithm to automatically adjust the belt velocities of the affected and sound sides and better biofeedback system with the sound leg information will be developed.

Kinematic walking analysis on a new vehicle “Tread-Walk” with active velocity control of treadmill belt

The importance on walking for health is growing in elder dominated society. We have been developing a new mobility “Tread-Walk 1 (TW-1)” controlled by walking movement. The device uses active treadmill velocity control, which allows the user to walk on the treadmill at any desired velocity. In this paper, the walking movements on the TW-1 were kinematically analyzed and compared with the walking movements on a traditional constant-velocity treadmill and on flat ground. The results showed that the walking pattern on the TW-1 was somewhat similar to that on a constant-velocity treadmill and on flat ground; however, the flexion angle of the hip joint and the dorsiflexion and plantaflexion angles of the ankle joint during TW-1 walking were larger. It also was shown that the foot applied a stronger kicking force to the belt at toe-off and the foot clearance on the TW-1 was larger than that on the constant-velocity treadmill and on flat ground. Therefore, the walking patterns in the swing and stance phase on the TW-1 are little different. However, the walking movements based on the TW-1 active belt control are valuable from the viewpoints of motion training.

Stable turning movement of a gait-controlled personal mobility “Tread-Walk 1”

We have been developing the new personal mobility aid, “Tread-Walk 1,” which is controlled by the drivers walking pace and amplifies walking velocity. The Tread-Walk 1 not only assists mobility but also helps the elderly maintain their physical abilities. However, the turning movement of the Tread-Walk is likely to be unstable, because the Tread-Walk is controlled by walking. In this paper, we focus on the stable turning movement of the Tread-Walk 1. First, we simulate the centrifugal force applied to the driver, which was not considered in our previous report. Then, a new algorithm to realize stable turning movement is established by measuring the drivers standing position. Finally, the developed algorithm for the turning movement of the Tread-Walk 1 is evaluated by measuring the handle grasping time. With the new algorithm, the Tread-Walk 1 turned steadily and driver did not lose their balance.