Real-time replanning using 3D environment for humanoid robot

Abstract

Abstract-In this paper, we illustrate experimentally an original real-time replanning scheme and architecture for humanoid robot reactive walking. Based on a dense set of actions, our approach uses a large panel of the humanoid robot capabilities and is particularly well suited for 3D collision avoidance. Indeed A-* approaches becomes difficult in such situation, thus the method demonstrated here relies on RRT. Combined with an approximation of the volume swept by the robot legs while walking, our method is able to cope with 3D obstacles while maintaining real-time computation. We experimentally validate our approach on the robot HRP-2.

Trajectory following for legged robots

“…This work has been motivated by the previous experimental setup described in [3]. In this paper, fast online replanning is used to handle environment changes.…”

“…Surprisingly, this issue has not been explicitly addressed in the literature concerning navigation for legged robots, although these machines are also prone to execution errors while moving. Previous experiments such as [3] illustrate how imprecise trajectory following on a humanoid robot can be. After executing a five meters long trajectory, the difference between the planned and real position can reach 0.4m.…”

Real-time 3D collision avoidance for biped robots

“…I our method uses ideas from [18]. Unlike [18], however, we limit the influence of the cost function to the six DoF of the swing-foot which are described by x F ∈ IR 6 . Since w can be written as w = [x T R , x T F ] T with the remaining coordinates x R ∈ IR (m−6) , we define a selection matrix S ∈ IR 6×m and a selection matrixS ∈ IR (m−6)×m…”

“…In this context, the ability of legged robots to step over or onto obstacles is mentioned as one of their main advantage over wheeled vehicles. While [2] proposes a method for autonomous navigation in unknown environments [3], [4], [5], [6], among others, present approaches for autonomous navigation in cluttered environments which are known in advance. They propose global footstep planners to reach goal positions while avoiding obstacles by using the ability to step over or onto obstacles.…”

Discrete Control Barrier Functions for Safety-Critical Control of Discrete Systems with Application to Bipedal Robot Navigation

“…In [20], a vision-based foot placement method is proposed, and implemented on the humanoid robot ASIMO, that searches over a discrete set of actions to avoid obstacles while reaching the goal. A real-time planning algorithm is presented in [3] that utilizes RRTs to search over dense, pre-computed swept volumes to avoid collision with moving 3D obstacles. Other work related to footstep placement planning include [19,6,5,18].…”