Autonomous evolution of dynamic gaits with two quadruped robots

A challenging task that must be accomplished for every legged robot is creating the walking and running behaviors needed for it to move. In this paper we describe our system for autonomously evolving dynamic gaits on two of Sony's quadruped robots. Our evolutionary algorithm runs on board the robot and uses the robot's sensors to compute the quality of a gait without assistance from the experimenter. First, we show the evolution of a pace and trot gait on the OPEN-R prototype robot. With the fastest gait, the robot moves at over 10 m/min, which is more than forty body-lengths/min. While these first gaits are somewhat sensitive to the robot and environment in which they are evolved, we then show the evolution of robust dynamic gaits, one of which is used on the ERS-110, the first consumer version of AIBO.     

Control of a quadruped standing jump over irregular terrain obstacles

In this paper the control of a quadruped standing jump over irregular terrain obstacles is investigated. Control strategies are developed for thetakeoff, flight and thelanding phases of a standing jump. Using a simplified planar model and the concept ofeffective linear momentum, simple feedforward leg force profiles are planned to remove the linear and angular momentum of the body during landing. Super real-time simulation, which involves predicting landing conditions based on faster than real-time simulation using the simplified model, is used to select leg touchdown angles for landing. Using the principle of symmetry, the leg forces during takeoff are derived from those predicted for landing. Leg motions are planned to maximize clearance during flight and stability during landing. Using these strategies, the quadruped is able to clear a variety of obstacles including isolated walls, terrain steps, and ditches. Simulation results are compared with experimental data of an animal jump from the literature.     

Generating high-speed dynamic running gaits in a quadruped robot using an evolutionary search

Over the past several decades, there has been a considerable interest in investigating high-speed dynamic gaits for legged robots. While much research has been published, both in the biomechanics and engineering fields regarding the analysis of these gaits, no single study has adequately characterized the dynamics of high-speed running as can be achieved in a realistic, yet simple, robotic system. The goal of this paper is to find the most energy-efficient, natural, and unconstrained gallop that can be achieved using a simulated quadrupedal robot with articulated legs, asymmetric mass distribution, and compliant legs. For comparison purposes, we also implement the bound and canter. The model used here is planar, although we will show that it captures much of the predominant dynamic characteristics observed in animals. While it is not our goal to prove anything about biological locomotion, the dynamic similarities between the gaits we produce and those found in animals does indicate a similar underlying dynamic mechanism. Thus, we will show that achieving natural, efficient high-speed locomotion is possible even with a fairly simple robotic system. To generate the high-speed gaits, we use an efficient evolutionary algorithm called set-based stochastic optimization. This algorithm finds open-loop control parameters to generate periodic trajectories for the body. Several alternative methods are tested to generate periodic trajectories for the legs. The combined solutions found by the evolutionary search and the periodic-leg methods, over a range of speeds up to 10.0 m/s, reveal "biological" characteristics that are emergent properties of the underlying gaits.     

Generation of discontinuous gaits for quadruped walking vehicles

Discontinuous gaits for walking machines have not yet been properly studied. Research has focused on the investigation, comprehension, and mathematical formulation of natural gaits. These gaits feature the fact that the body is in constant motion. The results have been significant, but they seem more adequate for animals than machines. On the other hand, discontinuous gaits, executed by animals under extreme conditions, exhibit excellent attributes for implementation in walking machines. This article presents a comparative study of continuous and discontinuous gaits with regard to their maximum achievable velocity and stability. Other aspects such as implementation in real machines, power requirements, and control under terrain difficulties are mentioned briefly. An elemental discontinuous gait is stated, and some variations on deriving crab and turning gaits are performed. Different methods for enlarging the achievable crab angle and improving stability are discussed for discontinuous crab gaits. A similar study is also done for turning gaits. (c) 1995 John Wiley & Sons, Inc.     

Continuous free-crab gaits for hexapod robots on a natural terrain with forbidden zones: An application to humanitarian demining

Autonomous robots are leaving the laboratories to master new outdoor applications, and walking robots in particular have already shown their potential advantages in these environments, especially on a natural terrain. Gait generation is the key to success in the negotiation of natural terrain with legged robots; however, most of the algorithms devised for hexapods have been tested under laboratory conditions. This paper presents the development of crab and turning gaits for hexapod robots on a natural terrain characterized by containing uneven ground and forbidden zones. The gaits we have developed rely on two empirical rules that derive three control modules that have been tested both under simulation and by experiment. The geometrical model of the SILO-6 walking robot has been used for simulation purposes, while the real SILO-6 walking robot has been used in the experiments. This robot was built as a mobile platform for a sensory system to detect and locate antipersonnel landmines in humanitarian demining missions.     

Policy gradient learning for quadruped soccer robots

In real-world robotic applications, many factors, both at low level (e.g., vision, motion control and behaviors) and at high level (e.g., plans and strategies) determine the quality of the robot performance. Consequently, fine tuning of the parameters, in the implementation of the basic functionalities, as well as in the strategic decisions, is a key issue in robot software development. In recent years, machine learning techniques have been successfully used to find optimal parameters for typical robotic functionalities. However, one major drawback of learning techniques is time consumption: in practical applications, methods designed for physical robots must be effective with small amounts of data. In this paper, we present a method for concurrent learning of best strategy and optimal parameters using policy gradient reinforcement learning algorithm. The results of our experimental work in a simulated environment and on a real robot show a very high convergence rate.     

The effect of motor action and different sensory modalities on terrain classification in a quadruped robot running with multiple gaits

Discriminating or classifying different terrains is an important ability for every autonomous mobile robot. A variety of sensors, preprocessing techniques, and algorithms in different robots were applied. However, little attention was paid to the way sensory data was generated and to the contribution of different sensory modalities. In this work, a quadruped robot traversing different grounds using a variety of gaits is used, equipped with a collection of proprioceptive (encoders on active, and passive compliant joints), inertial, and foot pressure sensors. The effect of different gaits on classification performance is assessed and it is demonstrated that separate terrain classifiers for each motor program should be employed. Furthermore, poor performance of randomly generated motor commands confirms the importance of coordinated behavior on sensory information structuring. The collection of sensors sensitive to active, “tactile”, terrain exploration proved effective. Among the individual modalities, encoders on passive compliant joints delivered best results.     

Study on effects of spinal joint for running quadruped robots

Intelligent Service Robotics - Most legged animals use their flexible body and supporting muscles to produce power for their locomotion, resulting in superior mobility and fast motions. In reality,...     

Generation of a continuous free gait for quadruped robot over rough terrains

Generating a robust gait is one of the most important factors to improve the adaptability of quadruped robots on rough terrains. This paper presents a new continuous free gait generation method for quadruped robots capable of walking on the rough terrain characterized by the uneven ground and forbidden areas. When walking with the proposed gait, the robot can effectively maintain its stability by using the Center of Gravity (COG) trajectory planning method. After analyzing the point cloud of rough terrain, the forbidden areas of the terrain can be obtained. Based on this analysis, an optimal foothold search strategy is presented to help quadruped robot to determine the optimum foothold for the swing foot automatically. In addition, the foot sequence determining method is proposed to improve the performance of robot. With the free gait proposed in this paper, quadruped robot can walk through the rough terrains automatically and successfully. The correctness and effectiveness of the proposed method is verified via simulations.     

Evolving gaits for hexapod robots using cyclic genetic algorithms

A major facet of multi-legged robot control is locomotion. Each leg must move in such a manner that it efficiently produces thrust and provides maximum support. The motion of all the legs must be coordinated so that they are working together to provide constant stability while propelling the robot forward. In this paper, we discuss the use of a cyclic genetic algorithm (CGA) to evolve control programs that produce gaits for actual hexapod robots. Tests done in simulation and verified on the actual robot show that the CGA successfully produces gaits for both fully capable and disabled robots.     

仿人机器人稳定步行控制研究

介绍了仿人机器人运动控制研究现状,通过对步行机器人稳定性判据ZMP分析,提出通过控制踝关节转动角度来调节ZMP的位置,以保证机器人行走的稳定性．根据模糊控制理论,设计出步行机器人踝关节二维模糊控制系统及模糊控制器．仿真结果表明步行机器人能够通过控制踝侧向关节的相对转动调节ZMP点的位置,实现机器人的稳定步行．     

A lattice-type reconfigurable robotic system for achieving gaits of chain-type robots

Traditional lattice-type reconfigurable robots can only achieve the flow-style locomotion with low efficiency. Since gaits of chain-type robots are proved to be efficient and practical, this paper presents a novel lattice distortion approach for lattice-type reconfigurable robots to achieve locomotion gaits of chain-type robots. Using this approach, the robotic system can be actuated by local lattice distortion to move as an ensemble. In this paper, a rule that makes the lattice distortion equivalent to joint rotation is presented firstly. Then, a kind of module structure is designed according to requirements of the lattice distortion. Finally, a motion planning for achieving locomotion is developed, which works well in physics-based simulations of completing a serpentine locomotion gait of a snake-like robot and a tripod gait of a hexapod robot.     

Terrain-evaluation-based motion planning for legged locomotion on irregular terrain

The path planning of legged locomotion is complex in that path generation is based on constraints not only from body motion, but also from leg motion. A general approach to path planning will fail in generating a feasible path for walking machines when facing the huge searching space of legged locomotion. In this paper, an effective method of path planning is introduced by virtue of terrain evaluation. It maps obstacles into the robot configuration space by evaluating the obstacles' influence on the legged locomotion. The evaluation produces an index of terrain, called terrain complexity, for path planning. Using potential-guided searching, the terrain with mapped obstacles is searched to generate a feasible path.     

Accurate tracking of legged robots on natural terrain

Statically stable walking locomotion research has focused mainly on robot design and gait generation. However, there is a need to expand robots’ capabilities so that walking machines can accomplish the kinds of real tasks for which they are eminently suited. Many such tasks demand trajectory tracking, but researchers have traditionally ignored this subject. This article focuses on the tracking of predefined trajectories with hexapod robots walking on natural terrain with forbidden zones. The method presented herein, which relies on gait algorithms defined elsewhere, describes certain localization strategies and control techniques that have been employed to follow trajectories accurately and have been implemented in a real walking hexapod. Several experimental examples are included to assess the proposed algorithms.     

Piecewise linear spine for speed–energy efficiency trade-off in quadruped robots

We compare the effects of linear and piecewise linear compliant spines on locomotion performance of quadruped robots in terms of energy efficiency and locomotion speed through a set of simulations and experiments. We first present a simple locomotion system that behaviorally resembles a bounding quadruped with flexible spine. Then, we show that robots with linear compliant spines have higher locomotion speed and lower cost of transportation in comparison with those with rigid spine. However, in linear case, optimal speed and minimum cost of transportation are attained at very different spine compliance values. Moreover, it is verified that fast and energy efficient locomotion can be achieved together when the spine flexibility is piecewise linear. Furthermore, it is shown that the robot with piecewise linear spine is more robust against changes in the load it carries. Superiority of piecewise linear spines over linear and rigid ones is additionally confirmed by simulating a quadruped robot in Webots and experiments on a crawling two-parts robot with flexible connection.     

基于改进蚁群算法的四足机器人步态规划

四足机器人关节众多、运动方式复杂,步态规划是四足机器人运动控制的基础。传统的算法多基于仿生原理,缺乏广泛适应性。 在建立运动学方程的基础上,提出了一种基于改进蚁群算法的步态规划算法。该算法利用了四足机器人4条腿运动的线性无关性,将步态规划问题转换为在四维空间里求取最长路径问题。仿真结果表明,该算法得出了满足约束条件的所有步态,最后通过机器人样机检验,验证了该算法求取结果的有效性和合理性。     

Whole-body motion planning and control of a quadruped robot for challenging terrain

Quadruped robots working in jungles, mountains or factories should be able to move through challenging scenarios. In this paper, we present a control framework for quadruped robots walking over rough terrain. The planner plans the trajectory of the robot's center of gravity by using the normalized energy stability criterion, which ensures that the robot is in the most stable state. A contact detection algorithm based on the probabilistic contact model is presented, which implements event-based state switching of the quadruped robot legs. And an on-line detection of contact force based on generalized momentum is also showed, which improves the accuracy of proprioceptive force estimation. A controller combining whole body control and virtual model control is proposed to achieve precise trajectory tracking and active compliance with environment interaction. Without any knowledge of the environment, the experiments of the quadruped robot SDUQuad-144 climbs over significant obstacles such as 38 cm high steps and 22.5 cm high stairs are designed to verify the feasibility of the proposed method.     

Video stabilization algorithm for field robots in uneven terrain

Artificial Life and Robotics - In recent years, many countries including Japan are facing the problems of increasing old-age population and shortage of labor. This has increased the demands of...     

Collision free path-planning for cable-driven parallel robots

In this study, a path-planning method that has been developed for serial manipulators is adapted to cable-driven robots. The proposed method has two modes. The first one is active when the robot is far from an obstacle. In this mode, the robot moves toward the goal on a straight line. The second mode is active when the robot is near an obstacle. During this mode, the robot finds the best way to avoid the obstacle. Moreover, an algorithm is presented to detect the collision between the robot and the obstacle. A similar algorithm is also presented to avoid the collision of the cables with an obstacle. Some simulation results are shown, which are then validated experimentally using a built 4-cable-driven parallel manipulator. Although the path obtained between the initial and final poses may not be the shortest possible one, it guarantees finding a path, when it exists, no matter how cluttered the environment is.     

Self‐supervised learning to visually detect terrain surfaces for autonomous robots operating in forested terrain

Autonomous robotic navigation in forested environments is difficult because of the highly variable appearance and geometric properties of the terrain. In most navigation systems, researchers assume a priori knowledge of the terrain appearance properties, geometric properties, or both. In forest environments, vegetation such as trees, shrubs, and bushes has appearance and geometric properties that vary with change of seasons, vegetation age, and vegetation species. In addition, in forested environments the terrain surface is often rough, sloped, and/or covered with a surface layer of grass, vegetation, or snow. The complexity of the forest environment presents difficult challenges for autonomous navigation systems. In this paper, a self‐supervised sensing approach is introduced that attempts to robustly identify a drivable terrain surface for robots operating in forested terrain. The sensing system employs both LIDAR and vision sensor data. There are three main stages in the system: feature learning, feature training, and terrain prediction. In the feature learning stage, 3D range points from LIDAR are analyzed to obtain an estimate of the ground surface location. In the feature training stage, the ground surface estimate is used to train a visual classifier to discriminate between ground and nonground regions of the image. In the prediction stage, the ground surface location can be estimated at high frequency solely from vision sensor data. © 2012 Wiley Periodicals, Inc.