Long Wang

Passive dynamic walking with flat feet and ankle compliance

This paper presents a bipedal locomotion model for passive dynamic walking with flat feet and compliant ankles. The two-dimensional seven-link model extends the simplest walking model with the addition of hip actuation, knee joints, flat feet and torsional springs based compliance on ankle joints, concerning heel-strike and toe-strike transitions, to achieve adaptive bipedal locomotion on level ground with controllable walking speed. We investigate the effects of foot geometric parameters and ankles stiffness on bipedal walking. The model achieves satisfactory walking results not only on even ground but also on uneven terrain with no active control and on different walking velocities. In addition, from the view of stability, there is an optimal foot-ankle ratio of the passivity-based walker. The results can be used to explore further understanding of bipedal walking, and help the design of future intelligent ankle-foot prosthesis and passivity-based robot prototypes towards more practical uses.

Formation Control of Heterogeneous Multi-Robot Systems

Abstract In this paper, a new position feedback based formation control method for heterogeneous multi-robot teams is presented and evaluated. The formation behaviors are integrated with dynamic reference object based collaborative navigation and efficient obstacle avoidance to maintain and change formation real-time. This method is computationally efficient and easy to coordinate in heterogeneous systems. The time to formalize and switch specified formation patterns can be controlled by adjusting the position feedback parameter. Satisfactory experimental results are obtained in simulation and real heterogeneous multi-robot system which consists of autonomous vehicles and legged robots.

Autonomous Evolution of High-Speed Quadruped Gaits Using Particle Swarm Optimization

This paper presents a novel evolutionary computation approach to optimize fast forward gaits for a quadruped robot with three motor-driven joints on each limb. Our learning approach uses Particle Swarm Optimization to search for a set of parameters automatically aiming to develop the fastest gait that an actual quadruped robot can possibly achieve, based on the concept of parameterized representation for quadruped gaits. In addition, we analyze the computational cost of Particle Swarm Optimization taking the memory requirements and processing limitation into consideration. Real robot experiments show that the evolutionary approach is effective in developing quadruped gaits. Satisfactory results are obtained in less than an hour by the autonomous learning process, which starts with randomly generated parameters instead of any hand-tuned parameters.

Learning from Human Cognition: Collaborative Localization for Vision-based Autonomous Robots

This paper presents a novel approach for a group of vision-based autonomous robots to localize in dynamic environments. We propose a hybrid system method for localization consisted of on-line and off-line subsystems inspired by human cognition. For the on-line subsystem, we use the landmark based Markov localization method to estimate the position. When the robot does not update the probability of current position through landmarks for a certain period, we use the off-line experience subsystem to help. In addition to the hybrid system for individual localization, we propose a method of dynamic reference object for collaborative localization. By using this method, an autonomous robot can estimate and correct its position perception more accurately and effectively, taking the odometry error and other negative influence into consideration. Satisfactory experimental results are obtained in the RoboCup Four-Legged League environment

Ground Contact Angle in Bipedal Locomotion towards Passive Dynamic Walking and Running

In this paper, we present the role of ground contact angle in passive dynamic walking and running with analysis by two limb models from an energetic view. First, a massless limb model is proposed which introduces locomotion cost of running as a function of speed, limb length, ground contact angle and excursion angle. Furthermore, a mass-added limb model is created taking the limb mass into account during walking and running to estimate the energetic cost of limb swing. Our models indicate that ground contact angle is relevant to the force production which is necessary to swing the limb. To estimate the importance of ground contact angle in bipedal locomotion during walking and running, we create a bipedal robot with the wheel chair pushing task. Only one motor is used to drive each limb to perform two dimensional movements. Passive movements of the foot and knee rely on gravity and on the elasticity of springs on the limbs. The relationship between ground contact angle and energetic cost in bipedal locomotion is justified by experiments in different gaits.

Let Robots Play Soccer under More Natural Conditions: Experience-Based Collaborative Localization in Four-Legged League

This paper presents an experience-based collaborative approach for a group of autonomous robots to localize in asymmetric, dynamic environments. To help robots play soccer under more natural conditions, we propose a Markov localization based hybrid method with integration of environment experience construction and dynamic reference object based multi-robot localization. By using this method, the robot can estimate and correct its position perception more accurately and effectively among a group of autonomous robots, taking the odometry error and other negative influence into consideration. Satisfactory results are obtained in the RoboCup Four-Legged League environment.

Integrated Heterogeneous Multi-Robot System for Collaborative Navigation

In this paper, we present an integrated heterogeneous multi-robot system that considers the problem of how heterogeneous autonomous robots, including wheeled robots and four-legged robots, which are different both in hardware and software, can collaborate to navigate in an unknown corridor environment. The navigation tasks involve detecting environment features, self-localizing, performing the planned navigation actions. On the one hand, a time- variable limit cycle based method is created to perform real-time obstacle avoidance. On the other hand, we propose a dynamic reference object based collaborative approach for self localization. Satisfactory experimental results are obtained in real robot experiments.

Collaborative Localization and Gait Optimization of SharPKUngfu Team

In this chapter, we introduce the recent progress of sharPKUngfu Team which participates in the RoboCup Four-Legged League since 2004. sharPKUngfu Team is a robot soccer team from Peking University, China. In July 2005, we got the third place in the RoboCup China Open. In June 2006, our sharPKUngfu Team has participated in the technical challenge of RoboCup 2006. In this event, our Medal Awarding challenge got the eighth place in the Open Challenge. In October 2006, we got the champion in the RoboCup China Open 2006, both in soccer competition and technical challenge. In July 2007, we participate in the RoboCup 2007 and got the fourth place in the technical challenge. Our research in robot soccer focuses on robot vision, multi-robot cooperation strategy, collaborative localization in dynamic environment, quadruped gaits optimization and intelligent behavior. We focus this chapter on localization and gait optimization which are the fundamental parts in soccer robotics. Recently, we successfully apply self-learning image-retrieval approach and collaboration in self localization in robot soccer. This improvement eliminates the problems of image-retrieval method and collaboration mentioned in previous research. By using this approach, robots can play soccer under more natural conditions towards real human soccer environment. We organize the localization part as follows. At first, a brief overview of current self-localization approaches is presented. Secondly, we introduce the human cognition inspired localization with self-learning experience. Specific algorithms for image features collection and self-learning process are described. Then, the dynamic reference object based method for collaborative localization is demonstrated in detail. Experimental results in real robot soccer are shown in the end. We also discuss current challenges and future works of localization in soccer robotics. How to get high-speed walking and running gaits is another problem in soccer robotics. Different to existing literature which uses Genetic Algorithms (GA) based gait optimization methods, we present the implementation of Particle Swarm Optimization (PSO) in generating high-speed gaits for a quadruped robot, specifically the Aibo, which is the commercial robot made in Sony. PSO has been proven to be effective in solving many global optimization problems and in some areas outperform many other optimization approaches including Genetic Algorithms. In this part, at first, we overview the basic PSO and Adaptive PSO (APSO) with comparison to other optimization approaches. After that, with the

Collaborative Localization Based Formation Control of Multiple Quadruped Robots

In this article, we present a new formation control method based on collaborative localization. Due to the accurate position estimation, the proposed approach is computationally light and well-performing in switching between different formations of real-world multi-robot systems. Satisfactory experimental results of multi-robot formation control are obtained in the RoboCup environment.

Fuzzy logic based body state estimation in a bipedal robot with passive dynamic gaits

In this paper, we present a hybrid body state estimator for the locomotion stability control of a bipedal robot with passive dynamic gaits. The hybrid system is based on the sensor data of a three-axis accelerometer and a digital compass which provides independent measurements of the bipedal locomotion and accurate sources of the disturbance information during passive dynamic walking and running. In addition, a fuzzy logic based algorithm is implemented in the system which performs feedback control. The body state estimator may play an important role not only in modeling and system stability but also in real-time control. Satisfactory results are obtained in a test platform that we create to evaluate bipedal locomotion with different gaits.