Showing papers on "Obstacle avoidance published in 2005"

Off-Road Obstacle Avoidance through End-to-End Learning

Abstract: We describe a vision-based obstacle avoidance system for off-road mobile robots. The system is trained from end to end to map raw input images to steering angles. It is trained in supervised mode to predict the steering angles provided by a human driver during training runs collected in a wide variety of terrains, weather conditions, lighting conditions, and obstacle types. The robot is a 50cm off-road truck, with two forward-pointing wireless color cameras. A remote computer processes the video and controls the robot via radio. The learning system is a large 6-layer convolutional network whose input is a single left/right pair of unprocessed low-resolution images. The robot exhibits an excellent ability to detect obstacles and navigate around them in real time at speeds of 2 m/s.

High speed obstacle avoidance using monocular vision and reinforcement learning

Abstract: We consider the task of driving a remote control car at high speeds through unstructured outdoor environments. We present an approach in which supervised learning is first used to estimate depths from single monocular images. The learning algorithm can be trained either on real camera images labeled with ground-truth distances to the closest obstacles, or on a training set consisting of synthetic graphics images. The resulting algorithm is able to learn monocular vision cues that accurately estimate the relative depths of obstacles in a scene. Reinforcement learning/policy search is then applied within a simulator that renders synthetic scenes. This learns a control policy that selects a steering direction as a function of the vision system's output. We present results evaluating the predictive ability of the algorithm both on held out test data, and in actual autonomous driving experiments.

A convergent dynamic window approach to obstacle avoidance

Abstract: The dynamic window approach (DWA) is a well-known navigation scheme developed by Fox et al. and extended by Brock and Khatib. It is safe by construction, and has been shown to perform very efficiently in experimental setups. However, one can construct examples where the proposed scheme fails to attain the goal configuration. What has been lacking is a theoretical treatment of the algorithm's convergence properties. Here we present such a treatment by merging the ideas of the DWA with the convergent, but less performance-oriented, scheme suggested by Rimon and Koditschek. Viewing the DWA as a model predictive control (MPC) method and using the control Lyapunov function (CLF) framework of Rimon and Koditschek, we draw inspiration from an MPC/CLF framework put forth by Primbs to propose a version of the DWA that is tractable and convergent.

6D-vision: fusion of stereo and motion for robust environment perception

Abstract: Obstacle avoidance is one of the most important challenges for mobile robots as well as future vision based driver assistance systems. This task requires a precise extraction of depth and the robust and fast detection of moving objects. In order to reach these goals, this paper considers vision as a process in space and time. It presents a powerful fusion of depth and motion information for image sequences taken from a moving observer. 3D-position and 3D-motion for a large number of image points are estimated simultaneously by means of Kalman-Filters. There is no need of prior error-prone segmentation. Thus, one gets a rich 6D representation that allows the detection of moving obstacles even in the presence of partial occlusion of foreground or background.

Motion Planning for Humanoid Robots

Abstract: Humanoid robotics hardware and control techniques have advanced rapidly during the last five years. Presently, several companies have announced the commercial availability of various humanoid robot prototypes. In order to improve the autonomy and overall functionality of these robots, reliable sensors, safety mechanisms, and general integrated software tools and techniques are needed. We believe that the development of practical motion planning algorithms and obstacle avoidance software for humanoid robots represents an important enabling technology. This paper gives an overview of some of our recent efforts to develop motion planning methods for humanoid robots for application tasks involving navigation, object grasping and manipulation, footstep placement, and dynamically-stable full-body motions. We show experimental results obtained by implementations running within a simulation environment as well as on actual humanoid robot hardware.

Vision-guided humanoid footstep planning for dynamic environments

Abstract: Despite the stable walking capabilities of modern biped humanoid robots, their ability to autonomously and safely navigate obstacle-filled, unpredictable environments has so far been limited We present an approach to autonomous humanoid walking that combines vision-based sensing with a footstep planner, allowing the robot to navigate toward a desired goal position while avoiding obstacles An environment map including the robot, goal, and obstacle locations is built in real-time from vision The footstep planner then computes an optimal sequence of footstep locations within a time-limited planning horizon Footstep plans are reused and only partially recomputed as the environment changes during the walking sequence In our experiments, combining real-time vision with plan reuse has allowed a Honda ASIMO humanoid robot to autonomously traverse dynamic environments containing unpredictably moving obstacles

Iterative MILP Methods for Vehicle Control Problems

Abstract: Mixed integer linear programming (MILP) is a powerful tool for planning and control problems because of its modeling capability and the availability of good solvers. However, for large models, MILP methods suffer computationally. In this paper, we present iterative MILP algorithms that address this issue. We consider trajectory generation problems with obstacle avoidance requirements and minimum time trajectory generation problems. The algorithms use fewer binary variables than standard MILP methods and require less computational effort.

Autonomous Exploration in Unknown Urban Environments for Unmanned Aerial Vehicles

Abstract: §In this paper, we present an autonomous exploration method for unmanned aerial vehicles in unknown urban environment. We address two major aspects of explorationgathering information about the surroundings and avoiding obstacles in the flight path- by building local obstacle maps and solving for confli ct-free trajectory using model predictive control (MPC) framework. For obstacle sensing, an onboard laser scanner is used to detect nearby objects around the vehicle. An MPC algorithm with a cost function that penalizes the proximity to the nearest obstacle replans the fligh t path in real-time. The adjusted trajectory is sent to the position tracking layer in the UAV a vionics. The proposed approach is implemented on Berkeley rotorcraft UAVs and successfully tested in a series of flights in urban obstacle setup.

Haptic Control of a Mobile Robot: A User Study

Abstract: We address the problem of teleoperating a mobile robot using shared autonomy: an on-board controller performs obstacle avoidance while the operator uses the manipulandum of a haptic probe to designate the desired speed and rate of turn. Sensors on the robot are used to measure obstacle range information. We describe a strategy to convert such range information into forces, which are reflected to the operator's hand, via the haptic probe. This haptic information provides feedback to the operator in addition to imagery from a front-facing camera mounted on the mobile robot. Extensive experiments with a user population show that the added haptic feedback significantly improves operator performance in several ways (reduced collisions, increased minimum distance between the robot and obstacles) without a significant increase in navigation time.

Obstacle Detection for Small Autonomous Aircraft Using Sky Segmentation

Abstract: A vision-based obstacle detection system for small unmanned aerial vehicles (UAVs) is presented. Obstacles are detected by segmenting the image into sky and non-sky regions and treating the non-sky regions as obstacles. The feasibility of this approach is demonstrated by using the vision output to steer a small unmanned aircraft to fly towards an obstacle. The experiment was first verified in a hardware in the loop (HIL) simulation and then successfully implemented on a small modified remote control plane using a large inflatable balloon as the obstacle.

Probabilistic aggregation strategies in swarm robotic systems

Abstract: In this study, a systematic analysis of probabilistic aggregation strategies in swarm robotic systems is presented. A generic aggregation behavior is proposed as a combination of four basic behaviors: obstacle avoidance, approach, repel, and wait. The latter three basic behaviors are combined using a three-state finite state machine with two probabilistic transitions among them. Two different metrics were used to compare performance of strategies. Through systematic experiments, how the aggregation performance, as measured by these two metrics, change 1) with transition probabilities, 2) with number of simulation steps, and 3) with arena size, is studied.

Navigating a uav with obstacle avoidance algorithms

Abstract: Methods, systems, and computer program products are provided for navigating a UAV that include piloting the UAV, under control of a navigation computer, in accordance with a navigation algorithm. While piloting the UAV, embodiments include reading from the GPS receiver a sequence of GPS data, anticipating a future position of the UAV, identifying an obstacle in dependence upon the future position, selecting an obstacle avoidance algorithm, and piloting the UAV in accordance with an obstacle avoidance algorithm. Identifying an obstacle in dependence upon the future position may include comprises retrieving obstacle data from a database in dependence the future position. Identifying an obstacle in dependence upon the future position may also include depicting an anticipated flight of the UAV with 3D computer graphics in dependence upon the future position and identifying an obstacle in dependence upon the depiction of the anticipated flight.

An overview of physicomimetics

Abstract: This paper provides an overview of our framework, called physicomimetics, for the distributed control of swarms of robots. We focus on robotic behaviors that are similar to those shown by solids, liquids, and gases. Solid formations are useful for distributed sensing tasks, while liquids are for obstacle avoidance tasks. Gases are handy for coverage tasks, such as surveillance and sweeping. Theoretical analyses are provided that allow us to reliably control these behaviors. Finally, our implementation on seven robots is summarized.

Passive bilateral control and tool dynamics rendering for nonlinear mechanical teleoperators

Abstract: We propose a passive bilateral teleoperation control law for a pair of n-degree-of-freedom (DOF) nonlinear robotic systems. The control law ensures energetic passivity of the closed-loop teleoperator with power scaling, coordinates motions of the master and slave robots, and installs useful task-specific dynamics for inertia scaling, motion guidance, and obstacle avoidance. Consequently, the closed-loop teleoperator behaves like a common passive mechanical tool. A key innovation is the passive decomposition, which decomposes the 2n-DOF nonlinear teleoperator dynamics into two robot-like systems without violating passivity: an n-DOF shape system representing the master-slave position coordination aspect, and an n-DOF locked system representing the dynamics of the coordinated teleoperator. The master-slave position coordination is then achieved by regulating the shape system, while programmable apparent inertia of the coordinated teleoperator is achieved by scaling the inertia of the locked system. To achieve this perfect coordination and inertia scaling, the proposed control law measures and compensates for environment and human forcing. Passive velocity field control and artificial potential field control are used to implement guidance and obstacle avoidance for the coordinated teleoperator. The designed control is also implemented in an intrinsically passive negative semidefinite structure to ensure energetic passivity of the closed-loop teleoperator, even in the presence of parametric model uncertainties and inaccurate force sensing. Experiments are performed to validate the properties of the proposed control framework.

Toward 30-gram Autonomous Indoor Aircraft: Vision-based Obstacle Avoidance and Altitude Control

Abstract: We aim at developing autonomous micro-flyers capable of navigating within houses or small built environments. The severe weight and energy constraints of indoor flying platforms led us to take inspiration from flying insects for the selection of sensors, signal processing, and behaviors. This paper presents the control strategies enabling obstacle avoidance and altitude control using only optic flow and gyroscopic information. For experimental convenience, the control strategies are first implemented and tested separately on a small wheeled robot featuring the same hardware as the targeted aircraft. The obstacle avoidance system is then transferred to a 30-gram aircraft, which demonstrates autonomous steering within a square textured arena.

Reinforcement Learning Neural Network to the Problem of Autonomous Mobile Robot Obstacle Avoidance

Abstract: An approach to the problem of autonomous mobile robot obstacle avoidance using reinforcement learning neural network is proposed in this paper. Q-learning is one kind of reinforcement learning method that is similar to dynamic programming and the neural network has a powerful ability to store the values. We integrate these two methods with the aim to ensure autonomous robot behavior in complicated unpredictable environment. The simulation results show that the simulated robot using the reinforcement learning neural network can enhance its learning ability obviously and can finish the given task in a complex environment.

Static and Dynamic Obstacle Avoidance for Miniature Air Vehicles

Abstract: Small unmanned air vehicles are limited in sensor weight and power such that detection and avoidance of unknown obstacles during flight is difficult. This paper presents a low power low weight method of detection using a laser range finder. In addition, a rapidly-exploring random tree algorithm to generate waypoint paths around obstacles known a priori is presented, and a dynamic geometric algorithm to generate paths around detected obstacles is derived. The algorithms are demonstrated in simulation and in flight tests on a fixed-wing miniature air vehicle (MAV). Index Words - Obstacle avoidance, waypoint path planning, rapidly exploring random tree, unmanned air vehicles, miniature air vehicles.

Real-Time Hierarchical POMDPs for Autonomous Robot Navigation

Abstract: This paper proposes a new hierarchical formulation of POMDPs for autonomous robot navigation that can be solved in real-time, and is memory efficient. It will be referred to in this paper as the Robot Navigation-Hierarchical POMDP (RN-HPOMDP). The RN-HPOMDP is utilized as a unified framework for autonomous robot navigation in dynamic environments. As such, it is used for localization, planning and local obstacle avoidance. Hence, the RN-HPOMDP decides at each time step the actions the robot should execute, without the intervention of any other external module for obstacle avoidance or localization. Our approach employs state space and action space hierarchy, and can effectively model large environments at a fine resolution. Finally, the notion of the reference POMDP is introduced. The latter holds all the information regarding motion and sensor uncertainty, which makes the proposed hierarchical structure memory efficient and enables fast learning. The RN-HPOMDP has been experimentally validated in real dynamic environments.

Obstacle avoidance of a mobile robot using hybrid learning approach

Abstract: in this paper, a hybrid learning approach for obstacle avoidance of a mobile robot is presented. the key features of the approach are, firstly, innate hardwired behaviors which are used to bootstrap learning in the mobile robot system. a neuro-fuzzy controller is developed from a pre-wired or innate controller based on supervised learning in a simulation environment. the fuzzy inference system has been constructed based on the generalized dynamic fuzzy neural networks learning algorithm of Wu and Er, whereby structure and parameters identification are carried out automatically and simultaneously. Secondly, the neuro-fuzzy controller is capable of re-adapting in a new environment. After carrying out the learning phase on a simulated robot, the controller is implemented on a real robot. A reinforcement learning method based on the fuzzy actor-critic learning algorithm is employed so that the system can re-adapt to a new environment without human intervention. In this phase, the structure of the fuzzy inference system and the parameters of the antecedent parts of fuzzy rules are frozen, and reinforcement learning is applied to further tune the parameters in the consequent parts of the fuzzy rules. Through the hybrid learning approach, an efficient and compact neuro-fuzzy system is generated for obstacle avoidance of a mobile robot in the real world.

Flocking with Obstacle Avoidance: A New Distributed Coordination Algorithm Based on Voronoi Partitions

Abstract: A new distributed coordination algorithm for multi-vehicle systems is presented in this paper. The algorithm combines a particular choice of navigation function with Voronoi partitions. This results not only in obstacle avoidance and motion to the goal, but also in a desirable geographical distribution of the vehicles. Our algorithm is decentralized in that each vehicle needs only to know the position of neigh boring vehicles, but no other inter-vehicle communication or centralized control are required. The algorithm gives a natural priority to safety, goal convergence, and formation keeping, in that (1) collision avoidance is guaranteed under all circumstances, (2) the vehicles will move toward the goal as long as a given optimization problem is feasible, and (3) if prior criteria admit, the vehicles tend to a desirable lattice formation. These theoretical properties are discussed in the paper and the performance of the algorithm is illustrated in simulations with realistic models of twenty all-terrain vehicles. Planned experimental evaluation using customized miniature cars is also briefly described.

Navigation Among Moving Obstacles Using the NLVO: Principles and Applications to Intelligent Vehicles

Abstract: Vehicle navigation in dynamic environments is a challenging task, especially when the motion of the obstacles populating the environment is unknown beforehand and is updated at runtime. Traditional motion planning approaches are too slow to be applied in real-time to this problem, whereas reactive navigation methods have generally a too short look-ahead horizon. Recently, iterative planning has emerged as a promising approach, however, it does not explicitly take into account the movements of the obstacles. This paper presents a real-time motion planning approach, based on the concept of the Non-Linear Vobst (NLVO) (Shiller et al., 2001). Given a predicted environment, the NLVO models the set of velocities which lead to collisions with static and moving obstacles, and an estimation of the times-to-collision. At each controller iteration, an iterative A* motion planner evaluates the potential moves of the robot, based on the computed NLVO and the traveling time. Previous search results are reused to both minimize computation and maintain the global coherence of the solutions. We first review the concept of the NLVO, and then present the iterative planner. The planner is then applied to vehicle navigation and demonstrated in a complex traffic scenario.

Autonomous Terrain Mapping and Classification Using Hidden Markov Models

Abstract: This paper presents a new approach for terrain mapping and classification using mobile robots with 2D laser range finders. Our algorithm generates 3D terrain maps and classifies navigable and non-navigable regions on those maps using Hidden Markov models. The maps generated by our approach can be used for path planning, navigation, local obstacle avoidance, detection of changes in the terrain, and object recognition. We propose a map segmentation algorithm based on Markov Random Fields, which removes small errors in the classification. In order to validate our algorithms, we present experimental results using two robotic platforms.

A Floor and Obstacle Height Map for 3D Navigation of a Humanoid Robot

Abstract: With the development of biped robots, systems became able to navigate in a 3 dimensional world, walking up and down stairs, or climbing over small obstacles. We present a method for obtaining a labeled 2.5D grid map of the robot's surroundings. Each cell is marked either as floor or obstacle and contains a value telling the height of the floor or obstacle. Such height maps are useful for path planning and collision avoidance. The method uses a novel combination of a 3D occupancy grid for robust sensor data interpretation and a 2.5D height map for fine resolution floor values. We evaluate our approach using stereo vision on the humanoid robot QRIO and show the advantages over previous methods. Experimental results from navigation runs on an obstacle course demonstrate the ability of the method to generate detailed maps for autonomous navigation.

Haptic teleoperation of a mobile robot: a user study

Abstract: The problem of teleoperating a mobile robot using shared autonomy is addressed: An onboard controller performs close-range obstacle avoidance while the operator uses the manipulandum of a haptic probe to designate the desired speed and rate of turn. Sensors on the robot are used to measure obstacle-range information. A strategy to convert such range information into forces is described, which are reflected to the operator's hand via the haptic probe. This haptic information provides feedback to the operator in addition to imagery from a front-facing camera mounted on the mobile robot. Extensive experiments with a user population both in virtual and in real environments show that this added haptic feedback significantly improves operator performance, as well as presence, in several ways (reduced collisions, increased minimum distance between the robot and obstacles, etc.) without a significant increase in navigation time.

Underwater path planing using fast marching algorithms

Abstract: In this paper, new tools for obstacle avoidance and path planning for underwater vehicles are presented. The authors' technique, based on a level set formulation of the path planning problem, extracts optimal paths from complex and continuous environments in a complete and consistent manner. Fast marching algorithm is known to be efficient for finding cost optimal path in mobile robotics because of its reliability, precision, and simple implementation. Fast marching algorithm originally propagates a wave front to isotropically explore the space. We propose an anisotropic version of fast marching by adding directional constraints in a cost function to minimize. We then propose a path planning method able to deal with vectorial fields of force for the first time. Furthermore we explore the relation between the curvature of the optimal path and the cost function generated from scalar and vectorial constraints. This a priori knowledge of the influence of the environment on the final path's curvature allows us to propose a solution to make sure a path is reachable by the vehicle according to its kinematics. A multi-resolution scheme based on an adaptive mesh generation is eventually introduced to speed up the overall algorithm. Results are shown computed from real and simulated underwater environments.

Swarms for chemical plume tracing

Abstract: This paper presents a physics-based framework for managing distributed sensor networks of autonomous vehicles, e.g., robots, which self-organize into structured lattice arrangements using only local information. The vehicles remain in formation during obstacle avoidance and search for a chemical emitter that is actively ejecting a toxic chemical into the air. We discuss a new plume tracing algorithm, based on the principles of fluid physics, that outperforms the leading biomimetic competitors for this task.