Takahiro Kagawa

Gait pattern generation for a power-assist device of paraplegic gait

We address a gait pattern generation on a legged locomotor device (WPAL: Wearable Power-Assist Locomotor) for paraplegics. In the gait movement with WPAL, a backward falling is a considerable problem, and a foot-floor collision during a swing movement would induce a loss of balance. In addition, adjustability of the gait parameters, such as stride length, gait cycle and maximum hight of the toe clearance, would be required for an individual paraplegic according to the degrees of his disabilities and skills. In this paper, we propose a gait pattern generation method considering the requirements of the stability and the adjustability. First, the trajectories of toe position, horizontal hip position, and foot plantar angle are calculated using a minimum jerk trajectory with the constraints of the position and velocity at via points. Second, the desired trajectories of joint angles are determined from the calculated trajectories by inverse kinematic equations. We demonstrate that generated desired trajectories for various gait parameter values and boundary conditions were satisfied with the required stability conditions.

A human interface for stride control on a wearable robot

A legged locomotor device for paraplegics have been attempted to improve their ADL and to prevent some complications. A stride control of the system based on the users intension is important to coordinate the voluntary movements of the user and the assisted movements of the paralyzed legs. In this paper, we propose a human interface with a walker to control the stride length of a legged locomotor device. Assuming that a intended stride is equal to a distance of the preceding movement of the walker, we developed a human interface estimating the movement distance of the walker, where the distance is calculated by polynomial fitting for the acceleration of the movement. In this study, we examine the proposed human interface from the measurement experiments of gait movements, and report the following results: (1) estimation accuracy by polynomial fitting method, and (2) feasibility of the adjustment of stride length using the proposed method. These results suggest that the proposed human interface is effective to adjust the stride length of a legged locomotor device.

Necessary condition for forward progression in ballistic walking.

Ballistic walking requires an appropriate configuration of posture and velocity at toe-off to avoid backward falling. In this study, we investigated a determinant of the state of the body center of mass (COM) at the toe-off with regard to ballistic walking. We used an inverted pendulum model to represent ballistic trajectories and the necessary condition for forward progression by a simple relationship between the COM states (position and velocity) at toe-off. This condition was validated through a computer simulation of a 7-link musculoskeletal model and measurement experiments of human movements involving stepping and walking. The results of the model simulation were in good agreement with some of the results predicted by the inverted pendulum model. The measurement experiments of walking and stepping movements showed that most COM states at toe-off satisfied the condition for forward progression and the measured trajectories during single support phase were similar to the ballistic trajectories although humans are capable of walking in non-ballistic ways. These results suggested that the necessary condition for forward progression can predict the COM states at toe-off for efficient movement and for avoiding backward falling during single support phase.

Whole-Body Motion Planning for Humanoid Robots by Specifying Via-Points

We design a framework about the planning of whole body motion for humanoid robots. Motion planning with various constraints is essential to success the task. In this research, we propose a motion planning method corresponding to various conditions for achieving the task. We specify some via-points to deal with the conditions for target achievement depending on various constraints. Together with certain constraints including task accomplishment, the via-point representation plays a crucial role in the optimization process of our method. Furthermore, the via-points as the optimization parameters are related to some physical conditions. We applied this method to generate the kicking motion of a humanoid robot HOAP-3. We have confirmed that the robot was able to complete the task of kicking a ball over an obstacle into a goal in addition to changing conditions of the location of a ball. These results show that the proposed motion planning method using via-point representation can increase articulation of the motion.

Motor synergies for dampening hand vibration during human walking.

This study investigated the motion required to carry a cup filled with water without spilling it, which is a common human dexterous task. This task requires the individual to dampen hand vibration while walking. We hypothesize that a reduction in hand jerk and a constant cup angle are required to achieve this task. We measured movements while human subjects carried a cup with water (WW task) and with stones (WS task) using a three-dimensional position measurement system and then analyzed joint coordination. We empirically confirmed that the value of hand jerk and the variance in cup angle in the WW task were smaller than those in the WS task. We used uncontrolled manifold (UCM) analysis to quantify joint coordination corresponding to the motor synergy required to reduce the hand jerk and variance of the cup angle. UCM components, which did not affect the hand jerk and cup angle, were larger than orthogonal components, which directly affected the hand jerk and cup angle in the WW task. These results suggest that there is a coordinated control mechanism that reduces hand jerk and maintains a constant cup angle when carrying a cup filled with water without spilling it. In addition, we suggest that humans adopt a flexible and coordinated control strategy of allowing variance independent of the variables that should be controlled to achieve this dexterous task.

State-dependent corrective reactions for backward balance losses during human walking

We investigated corrective reactions for backward balance losses during walking. Several biomechanical studies have suggested that backward falling can be predicted from the horizontal position and velocity of the body center of mass (COM) related to the stance foot. Our hypothesis was that corrective reactions for backward balance losses depend on whether the body moves forward or backward after a perturbation. Using a split-belt treadmill, backward balance losses during walking were induced by rapid decreases of belt speed from 3.5 km/h to 2.5, 2.0, 1.5 and 1.0 km/h. We measured kinematic data and surface electromyography (EMG) during corrective reactions while walking on the treadmill. Phase portrait analysis of COM trajectories revealed that backward balance stability was decreased by the perturbations. When the perturbed belt speed was 1.0 km/h, the COM states at toe-off were significantly lower than the stability limit; a rapid touch-down of the swing foot posterior to the stance foot then occurred, and the gait rhythm was modulated so that the phase advanced. EMG recordings during perturbed steps revealed a bilateral response, including modulation of the swing leg during the recovery. For weaker perturbations, the swing foot placements were anterior to the stance foot and there was a phase delay. In contrast to the bilateral responses for stronger perturbations, unilateral EMG responses were observed for weaker perturbations. The differences in joint kinematics and EMG patterns in the unperturbed swing leg depended on the COM states at toe-off, suggesting the existence of different responses consisting of ongoing swing movements and rapid touch-down. Thus, we conclude that corrective reactions for backward balance losses are not only phase-dependent but also state-dependent. In addition, the control system for backward balance losses predicts the feasibility of forward progression and modulates swing movement and walking rhythm according to backward balance stability.

Optimization-Based Motion Planning in Joint Space for Walking Assistance With Wearable Robot

In this paper, we propose an alternative motion planning method for a wearable robot with a variable stride length and walking speed. Trajectories are planned in a joint space rather than a workspace to avoid an ill-posed problem with no solution in inverse kinematics, and to consider the joints range of motion, maximum velocity, foot clearance, and backward balance. The joint trajectories are represented by minimum jerk trajectories. Two via-points are assigned, and the parameters (angle and angular velocity) at the via-points are determined by applying an inverted pendulum model or optimization to satisfy the constraints. The fastest gait pattern generated by the proposed algorithm was twice as fast as the pattern generated by the workspace-based planning method. We confirmed that the fastest walking pattern of 0.36 m/s was feasible on a treadmill, and a walking pattern of 0.27 m/s was found for walking across the floor with a walker. Furthermore, the proposed method required approximately 65% of the electric power for the workspace-based method for the same walking speed and stride length. These results suggest that the proposed motion planning method is effective at generating a high-speed and efficient gait pattern for a wearable robot.

Change of a motor synergy for dampening hand vibration depending on a task difficulty

The present study investigated the relationship between the number of usable degrees of freedom (DOFs) and joint coordination during a human-dampening hand vibration task. Participants stood on a platform generating an anterior–posterior directional oscillation and held a water-filled cup. Their usable DOFs were changed under the following conditions of limb constraint: (1) no constraint; (2) ankle constrained; and (3) ankle–knee constrained. Kinematic whole-body data were recorded using a three-dimensional position measurement system. The jerk of each body part was evaluated as an index of oscillation intensity. To quantify joint coordination, an uncontrolled manifold (UCM) analysis was applied and the variance of joints related to hand jerk divided into two components: a UCM component that did not affect hand jerk and an orthogonal (ORT) component that directly affected hand jerk. The results showed that hand jerk when the task used a cup filled with water was significantly smaller than when a cup containing stones was used, regardless of limb constraint condition. Thus, participants dampened their hand vibration utilizing usable joint DOFs. According to UCM analysis, increasing the oscillation velocity and the decrease in usable DOFs by the limb constraints led to an increase of total variance of the joints and the UCM component, indicating that a synergy-dampening hand vibration was enhanced. These results show that the variance of usable joint DOFs is more fitted to the UCM subspace when the joints are varied by increasing the velocity and limb constraints and suggest that humans adopt enhanced synergies to achieve more difficult tasks.

Planning of kicking motion with via-point representation for humanoid robots

This paper describes planning of a whole body motion, such as kicking motion, for humanoid robots. Motion planning which is generally accompanied by various constraints is a key problem to achieve the task. In these constraints, the conditions for achieving the task are the most important component in the motion planning. In this research, we propose a method for motion planning with via-point representation to deal with various conditions for achieving the task. The via-point representation plays a crucial role in the reappearance of human movements. Our method deals with various constraints including the achievements for the task by conditions of the via-point parameterization in the optimization process. We applied this method to generate the kicking motion of a humanoid robot HOAP-3, and confirmed that the robot kicked a ball to roll on the ground and to fly in the air.

Efficient Planning of Humanoid Motions by Modifying Constraints

Abstract In this paper, we propose an effective planning method for whole-body motions of humanoid robots under various conditions for achieving the task. In motion planning, various constraints such as range of motion have to be considered. Specifically, it is important to maintain balance in whole-body motion. In order to be useful in an unpredictable environment, rapid planning is an essential problem. In this research, via-point representation is used for assigning sufficient conditions to deal with various constraints in the movement. The position, posture and velocity of the robot are constrained as a state of a via-point. In our algorithm, the feasible motions are planned by modifying via-points. Furthermore, we formulate the motion planning problem as a simple iterative method with a Linear Programming (LP) problem for efficiency of the motion planning. We have applied the method to generate the kicking motion of a HOAP-3 humanoid robot. We confirmed that the robot can successfully score a goal with various courses corresponding to changing conditions of the location of an obstacle. The computation time was less than two seconds. These results indicate that the proposed algorithm can achieve efficient motion planning.