Real-time pattern generation among obstacles for biped robots

Abstract

Abstract-In this paper we present a step-planner embedded in a framework which enables a humanoid robot to navigate among obstacles, exploiting its overall capacities. The system allows the robot to react to changes of user input in real-time while walking at reasonable speeds. The proposed method relies neither on external sensors nor on color coding or textured surfaces. The key idea is to use a fast collision model based on swept-sphere-volumes (SSVs) for real-time generation of collision-free footsteps and whole-body trajectories. Using a SSV-based 3D approximation in all control modules enables the robot to avoid collisions with itself and the environment. Obstacles are detected with an on-board RGB-D sensor while the robot navigates through an environment which is not known in advance. A step-planner reacts to high-level user commands like desired velocity and direction within less than a step. Instead of investigating only the footholds an articulated 3D approximation of the lower leg and the foot is considered to find feasible and optimal footstep locations. Additionally, it provides an initial solution for the swing-foot movement. Final, Collision-free swing-foot trajectories are created in real-time in the feedback control layer using all the foot's degrees of freedom. We validated this approach in experiments with our robot Lola.

Fast object approximation for real-time 3D obstacle avoidance with biped robots

“…This compact representation of a non-horizontal terrain is computationally very efficient, therefore it was used in most early works on autonomous navigation, as well as in our previous implementation [5]. In 2003 Kagami et al [7] used stereo cameras to create a height-map of the terrain, from which they could also extract walkable planar surfaces.…”

“…In the feedback control layer the Trajectory optimization takes into account collisions as well as selfcollisions by optimizing all the foot's degrees of freedom. Details are presented in [4], [5].…”

“…Email: daniel.wahrmann@tum.de strategy was initially presented in [4], where obstacle approximations were manually sent to the system and no perception was involved. It was extended in [5] to use a more general A*-based solution in a framework that admits perception information. It made use of a simple vision system based on height-maps, which was not presented.…”

“…In this work we present a new, more complex vision system which admits a full 3D representation of the environment, generates better and more complex approximations and can track moving objects. It is integrated with our planning system [5] to autonomously navigate in cluttered, dynamic scenarios. Its main characteristics are:…”

Real-time predictive kinematic evaluation and optimization for biped robots

“…foot-ground contact, which are critical in order to maintain balance. In this context, our objective is to further extend our framework for whole-body collision avoidance presented in [17], [18], [19], [20]. In 2014 [18] we proposed a local online optimization technique exploiting all the robot's swing-foot DoFs for stepping over arbitrarily shaped obstacles while avoiding collisions with the environment and self-collisions.…”