We propose a passive semi-autonomous control law of flippers to assist a crawler-type robot with four flippers running over rough terrain. In the proposed method, the robot presses its flippers against the ground and controls the pressing force so that the flippers passively adapt to terrain and avoid overturning. The operator sets a common target angle for all flippers, and the actual flipper angles are “semi-autonomously” determined through the interaction between the limited maximum torque of flippers and the reaction force from the terrain. Since shape adaptation to terrain is passive, there's no need for environmental sensing and no risk of malfunction due to environmental sensor noise. Furthermore, as all flippers are driven independently, the robot can maintain its posture in roll directions, where the risk of overturning is high. Verification using the robot FUHGA3 with the proposed control law confirmed high mobility over rough terrain with simple maneuvers.

In this study, we define the program scheduling problem and propose a scheduling method for the problem. The program scheduling problem deals with multiple projects that are interrelated in terms of precedence relationships and resource utilization. This problem is divided into two phases: the program scheduling phase and the project scheduling phase. In the program scheduling phase, we assume a situation where project information is progressively elaborated. Therefore, the start time and the amount of resource allocations for the executable projects are determined dynamically. In the project scheduling phase, the start times of project activities are determined, considering constraints of precedence and resource availability. The availability of resources corresponds to the amount of resources allocated during the program scheduling phase. Our proposed scheduling method divides the target problem into several decision-making problems, such as allocating resources to the project, calculating the project makespan based on the resource allocation, and determining the project start time and processing mode. The numerical experiment shows that our proposed method is effective through several simulations.