To maximize total operating time, an online minimum-energy translational and rotational velocity trajectory planning and control system is presented on a straight-line path for three-wheeled omnidirectional mobile robots (TOMRs). We suggest an efficient online trajectory planning algorithm, which minimizes a practical cost function as the energy that is drawn from batteries to motors, which is based on an accurate TOMR dynamic model that includes both the Coriolis force and the actuator motor dynamics. Using Pontryagin's minimum principle, we find the minimum-energy rotational velocity trajectory in analytic form. Then, the minimum-energy translational velocity trajectory is found using a novel algorithm that has a time complexity of O(n), which is based on a linearity condition on the state transition of TOMRs. Moreover, a trajectory control system is implemented using the resolved-acceleration control to validate the actual performance. Simulation and experiment results show that these minimum-energy trajectories can save energy up to 4.76% compared with the energy-efficient trapezoidal velocity profile and up to 5.29% compared with the loss-minimization control.

Training with visual and auditory biofeedback, in patients with stroke, improved balance ability and asymmetric posture. We developed a new biofeedback training device to prevent falls and improve balance ability in patients with stroke. This device corrects motion errors by collecting the pressure information of patients in real-time. This randomized crossover study aimed to investigate the effect of this biofeedback training on the static balance ability and weight distribution symmetry index in 24 patients with chronic stroke. Pressure sensor-based vibrotactile biofeedback, visual biofeedback providing posture information, and standing without biofeedback were randomly applied for 1 d each with 24 h washout intervals to minimize adaptation. The static balance ability was measured for each biofeedback training type, and the weight distribution symmetry index was calculated using the collected weight-bearing rate data. The static balance ability and weight distribution symmetry index differed significantly according to the type of biofeedback training used. Post-hoc analysis revealed significant differences in the order of newly developed vibrotactile biofeedback, visual biofeedback, and standing without biofeedback. These findings provide evidence that pressure sensor-based vibrotactile biofeedback improves static balance ability and weight support rates by proposing better intervention for patients with chronic stroke in the clinical environment.

A novel online path refinement method is suggested for mobile robots, which refines the given path, including several waypoints using sensor data with a low computational load. The method periodically plans a waypoint path from the current robot position to a certain waypoint on the given waypoint path. While the robot moves along the given path, the path is continuously refined to avoid collisions and local minima caused by unmapped obstacles that are not represented on the map but sensed by sensors. Also, if a shortcut is detected while the robot is moving, the shortcut replaces the existing waypoint path. This online waypoint path refinement (OWPR) method is very practical in terms of computation time and memory size since it is performed regardless of the size of the workspace. To validate the actual performances of OWPR method, various simulations and experiments are conducted. As a result of the simulation and experiment, the maximum execution time of the OWPR algorithm is less than 10 ms, and the generated waypoint path shortened the overall travel distance without collision and falling to the local minima.