Surface Profile Determination from Multiple Sonar Data Using Morphological Processing

This paper presents a novel method for surface profile determination using multiple sensors. Our approach is based on morphological processing techniques to fuse the range data from multiple sensor returns in a manner that directly reveals the target surface profile. The method has the intrinsic ability of suppressing spurious readings due to noise, crosstalk, and higher-order reflections, as well as processing multiple reflections informatively. The approach taken is extremely flexible and robust, in addition to being simple and straightforward. It can deal with arbitrary numbers and configurations of sensors as well as synthetic arrays. The algorithm is verified both by simulations and experiments in the laboratory by processing real sonar data obtained from a mobile robot. The results are compared to those obtained from a more accurate structured-light system, which is, however, more complex and expensive.

Motion primitives and 3D path planning for fast flight through a forest

This paper presents two families of motion primitives for enabling fast, agile flight through a dense obstacle field. The first family of primitives consists of a time-delay dependent 3D circular path between two points in space and the control inputs required to fly the path. In particular, the control inputs are calculated using algebraic equations which depend on the flight parameters and the location of the waypoint. Moreover, the transition between successive maneuver states, where each state is defined by a unique combination of constant control inputs, is modeled rigorously as an instantaneous switch between the two maneuver states following a time delay which is directly related to the agility of the robotic aircraft. The second family consists of aggressive turn-around (ATA) maneuvers which the robot uses to retreat from impenetrable pockets of obstacles. The ATA maneuver consists of an orchestrated sequence of three sets of constant control inputs. The duration of the first segment is used to optimize the ATA for the spatial constraints imposed by the turning volume. The motion primitives are validated experimentally and implemented in a simulated receding horizon control (RHC)-based motion planner. The paper concludes with inverse-design pointers derived from the primitives.

Mobile Robot Localization Using Fuzzy Segments

This paper presents the development of a framework based on fuzzy logic for multi-sensor fusion and localization in indoor environments. Such a framework makes use of fuzzy segments to represent uncertain location information from different sources of information. Fuzzy reasoning, based on similarity interpretation from fuzzy logic, is then used to fuse the sensory information represented as fuzzy segments. This approach makes it possible to fuse vague and imprecise information from different sensors at the feature level instead of fusing raw data directly from different sources of information. The resulting fuzzy segments are used to maintain a coherent representation of the environment around the robot. Such an uncertain representation is finally used to estimate the robot position. The proposed multi-sensor fusion localization approach has been validated with a mobile platform using different range sensors.

Neural network approaches to dynamic collision-free trajectory generation

In this paper, dynamic collision-free trajectory generation in a nonstationary environment is studied using biologically inspired neural network approaches. The proposed neural network is topologically organized, where the dynamics of each neuron is characterized by a shunting equation or an additive equation. The state space of the neural network can be either the Cartesian workspace or the joint space of multi-joint robot manipulators. There are only local lateral connections among neurons. The real-time optimal trajectory is generated through the dynamic activity landscape of the neural network without explicitly searching over the free space nor the collision paths, without explicitly optimizing any global cost functions, without any prior knowledge of the dynamic environment, and without any learning procedures. Therefore the model algorithm is computationally efficient. The stability of the neural network system is guaranteed by the existence of a Lyapunov function candidate. In addition, this model is not very sensitive to the model parameters. Several model variations are presented and the differences are discussed. As examples, the proposed models are applied to generate collision-free trajectories for a mobile robot to solve a maze-type of problem, to avoid concave U-shaped obstacles, to track a moving target and at the same to avoid varying obstacles, and to generate a trajectory for a two-link planar robot with two targets. The effectiveness and efficiency of the proposed approaches are demonstrated through simulation and comparison studies.

Range-only fuzzy Voronoi-enhanced localization of mobile robots in wireless sensor networks

Wireless Sensor Network (WSN) localization has shown a growing research interest, thanks to the expected proliferation of WSN applications. This work is focused on indoor localization of a mobile robot in a WSN using only inter-node range measurements, which are estimated by radio frequency signal strength attenuation. These measurements are affected by different sources of uncertainty that make them highly noisy and unreliable. The proposed approach makes use of fuzzy logic for modeling and dealing with such uncertain information. Besides, the position estimation is enhanced using a rough description of indoor environment. The experiments show that the proposed localization approach (i) is fault-tolerant, (ii) results feasible in low-density WSNs, and (iii) provides better position estimations than well-known localization methods when the position measurements are affected by high uncertainty.

CONCURRENT MAP BUILDING AND LOCALIZATION ON INDOOR DYNAMIC ENVIRONMENTS

A system that builds and maintains a dynamic map for a mobile robot is presented. A learning rule associated to each observed landmark is used to compute its robustness. The position of the robot during map construction is estimated by combining sensor readings, motion commands, and the current map state by means of an Extended Kalman Filter. The combination of landmark strength validation and Kalman filtering for map updating and robot position estimation allows for robust learning of moderately dynamic indoor environments.

Fuzzy mobile-robot positioning in intelligent spaces using wireless sensor networks

This work presents the development and experimental evaluation of a method based on fuzzy logic to locate mobile robots in an Intelligent Space using wireless sensor networks (WSNs). The problem consists of locating a mobile node using only inter-node range measurements, which are estimated by radio frequency signal strength attenuation. The sensor model of these measurements is very noisy and unreliable. The proposed method makes use of fuzzy logic for modeling and dealing with such uncertain information. Besides, the proposed approach is compared with a probabilistic technique showing that the fuzzy approach is able to handle highly uncertain situations that are difficult to manage by well-known localization methods.

Multiresolutional Fusion of Perceptions Applied to Robot Navigation

Visual perception-based autonomously moving virtual agent in virtual reality as a counterpart of an actual robot moving with a given dynamics is investigated. The visual perception is mathematically modelled as a probabilistic process obtaining and interpreting visual information from an environment. The perception obtained in the form of measurements in 2D is used for perceptual robot navigation. By means of this twofold gain is obtained; while the autonomous robot is navigated, it is equipped with some human-like behaviour, thereby dealing with complexity and environmental dynamics. The visual data is processed in a multiresolutional form via wavelet transform and optimally estimated via extended Kalman filtering in each resolution level and the outcomes are fused for improved estimation of the trajectory. The perceptual robotics experiments are carried out in virtual reality for the demonstration of the feasibility of the investigations in this domain. The computer experiments are carried out with perception measurement data, and the sensor/data fusion experiments are carried out by means of simulation. The improvement on the trajectory estimation by means of sensor/data fusion is demonstrated. The research is connected to building technological robotics, where some form of perceptual intelligence, like reaction to moving objects around, is required during operation.

An efficient dynamic system for real-time robot-path planning

This paper presents a simple yet efficient dynamic-programming (DP) shortest path algorithm for real-time collision-free robot-path planning applicable to situations in which targets and barriers are permitted to move. The algorithm works in real time and requires no prior knowledge of target or barrier movements. In the case that the barriers are stationary, this paper proves that this algorithm always results in the robot catching the target, provided it moves at a greater speed than the target, and the dynamic-system update frequency is sufficiently large. Like most robot-path-planning approaches, the environment is represented by a topologically organized map. Each grid point on the map has only local connections to its neighboring grid points from which it receives information in real time. The information stored at each point is a current estimate of the distance to the nearest target and the neighbor from which this distance was determined. Updating the distance estimate at each grid point is done using only the information gathered from the point's neighbors, that is, each point can be considered an independent processor, and the order in which grid points are updated is not determined based on global knowledge of the current distances at each point or the previous history of each point. The robot path is determined in real time completely from the information at the robot's current grid-point location. The computational effort to update each point is minimal, allowing for rapid propagation of the distance information outward along the grid from the target locations. In the static situation, where both the targets and the barriers do not move, this algorithm is a DP solution to the shortest path problem, but is restricted by lack of global knowledge. In this case, this paper proves that the dynamic system converges in a small number of iterations to a state where the minimal distance to a target is recorded at each grid point and shows that this robot-path-planning algorithm can be made to always choose an optimal path. The effectiveness of this algorithm is demonstrated through a number of simulations.

Training of a leaning agent for navigation--inspired by brain-machine interface

The design clue for the remote control of a mobile robot is inspired by the Talwar's brain-machine interface technology for remotely training and controlling rats. Our biologically inspired autonomous robot control consciousness-based architecture (CBA) is used for the remote control of a robot as a substitute for a rat. CBA is a developmental hierarchy model of the relationship between consciousness and behavior, including a training algorithm. This training algorithm computes a shortcut path to a goal using a cognitive map created based on behavior obstructions during a single successful trial. However, failures in reaching the goal due to errors of the vision and dead reckoning sensors require human intervention to improve autonomous navigation. A human operator remotely intervenes in autonomous behaviors in two ways: low-level intervention in reflexive actions and high-level ones in the cognitive map. Experiments are conducted to test CBA functions for intervention with a joystick for a Khepera robot navigating from the center of a square obstacle with an open side toward a goal. Their statistical results show that both human interventions, especially high-level ones, are effective in drastically improving the success rate of autonomous detours.

Qualitative map learning based on covisibility of objects

Autonomous map construction is one of the most fundamental and significant issues in intelligent mobile robot research. While a variety of map construction methods have been proposed, most require some quantitative measurements of the environment and a mechanism of precise self-localization. This paper proposes a novel map construction method using only qualitative information about "how often two objects are observed simultaneously." This method is based on heuristics--"closely located objects are likely to be seen simultaneously more often than distant objects" and a well-known multivariate data analysis technique-multidimensional scaling. A significant feature of this method is that it requires neither quantitative sensor measurements nor information about the robot's own position. Simulation and experimental results demonstrated that this method is sufficiently practical for capturing a qualitative spatial relationship among identifiable landmark objects rapidly.