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

Abstract

Abstract-In order to achieve fully autonomous humanoid navigation, environment perception must be both fast enough for real-time planning in dynamic environments and robust against previously unknown scenarios. We present an open source, flexible and efficient vision system that represents dynamic environments using simple geometries. Based only on onboard sensing and 3D point cloud processing, it approximates objects using swept-sphere-volumes while the robot is moving. It does not rely on color or any previous models or information. We demonstrate the viability of our approach by testing it on our human-sized biped robot Lola, which is able to avoid moving obstacles in real-time while walking at a set speed of 0.4m/s and performing whole-body collision avoidance.

Fast and Continuous Foothold Adaptation for Dynamic Locomotion Through CNNs

“…Despite their ability to perform locomotion tasks with onboard sensors only, most of the vision-based foothold selection strategies involve slow motions, mainly to provide enough time to complete the most costly operations such as image processing and optimization. An exception was shown by Wahrmann et al [19], where the acquisition of swept-spherevolumes allowed the biped robot Lola to avoid obstacles while moving, with no prior information about the environment. Nevertheless, this strategy was mainly demonstrated for single obstacle avoidance and self collision, and not for rough terrain.…”

Real-time predictive kinematic evaluation and optimization for biped robots

“…It works with a cycle time of 30 ms. For this reason changes of the perceived environment, e.g. moving obstacles, can be taken into account while the robot is walking [20]. (b) With each step an A*-search based…”

“…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.…”

“…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. The local optimization technique is combined with an A*-based footstep planner and an on-board vision system in order to close the gap between obstacle recognition, collision-free walking in 2D and whole-body 3D collision avoidance [19], [20]. To improve both robustness and performance of the robot's movements we propose to replace the previous heuristic approach by evaluating and optimizing the kinematics of the stepping movements.…”

“…The paper is organized as follows: In Sec. II we give a system overview of the experimental platform used in this work -the robot Lola -and a review of our framework for collision avoidance [17], [18], [19], [20]. We then present an online kinematic optimization method for robot movements in cluttered environments in Sec.…”