Catálogo de publicaciones - libros

This chapter starts with an overview and classification of robots: industrial robots, autonomous mobile robots, humanoid robots and educational robots. A specification is given of a generic educational robot used throughout the book: a small mobile robot with differential drive and horizontal and ground proximity sensors. A pseudocode is defined so that algorithms can be presented in a platform-independent manner. The chapter concludes with a detailed overview of the contents of the book.

Robots must sense their environment. Sensors use ultrasound, infrared and lasers to determine distance and angles. Cameras are essential for identifying objects in the environment. Important properties of sensors are their range, resolution, precision and accuracy. The response of a sensor may be linear, but if not calibration is performed so that values returned by the sensor can be interpreted as physical quantities.

The behavior of a robot is reactive if values returned by the sensors directly affect the actuators. Two families of reactive behavior are presented. Braitenberg vehicles are very simple robots that demonstrate relatively complex behavior. A robot with ground sensors can follow a line on a surface and thus navigate a known environment such as a warehouse. Three algorithms are presented: an algorithm that uses two ground sensors, an algorithm that uses only one ground sensor on a line with a gradient and an algorithm that follows the edge of a line.

Robots have embedded computers with memory that can be used to store the current state of an algorithm. Finite state machines specify the conditions under which the state of the robot changes and the actions taken when the state changes. Finite state machines are demonstrated first by Braitenberg vehicles and then by an algorithm that causes the robot to search for an object and then approach it.

The focus in this book is on mobile robots that move on a surface. When the robot moves for a period of time its new position can be determined by odometry: integrating the velocity of the robot over the period of its motion to obtain distance or integrating the acceleration to get velocity and integrating again to obtain distance. If the robot changes its heading as it moves, trigonometry is needed to compute the new position. Odometry is subject to errors caused by uncertainty in the components of the robot and unevenness of the surface. Wheel encoders enable more accurate odometry. Inertial navigation systems perform odometry based on accurate measurement of linear and angular acceleration. The degrees of freedom (DOF) of a system is the number dimensions through which it can move. The number of actuators of the robot can be more or less than the DOF. The DOF can be different from the degrees of mobility (DOM), which are the DOF that can be directly accessed. The concept of holonomic motion relates the DOF and the DOM.

Robots use feedback control algorithms which compute commands to the actuators based upon the error between the current state of the robot and its target state. The magnitudes of the commands can be proportional to the error, obtained by integrating or differentiating the error, or some combination of these functions. The goal is to reach the target state, to reach it quickly and to avoid instability such as oscillations. The performance of a control system depends on parameters called gains, which can be determined by experimentation.

A mobile robot navigating to a goal will encounter obstacles that must be avoided. Obstacle avoidance is presented in the context of wall following: moving around the wall of the obstacle until it no longer prevents access to the goal. Three algorithms are presented: simple wall following which fails in the presence of multiple obstacles, wall following with direction that also fails for certain obstacles and the Pledge algorithm which can avoid these obstacles. To demonstrate finding a path to a goal, an algorithm is presented that is based upon the behavior of a colony of ants searching for a food source even though they do not know its location. A probabilistic model explains the success of the algorithm.

A robot must perform localization, that is, it must know its own position in the environment. It can compute its position relative to landmarks whose position is known using the techniques of classical land surveying. Accurate localization can be achieved using the satellite-based global positioning system (GPS). Probabilistic localization is used in environments where GPS cannot be used, for example, inside a building. Given a map of the environment the robot computes its most likely location based upon data from its sensors. The probabilistic computations can handle uncertainty in the sensors and in the motion of the robot.

A robot uses a map to localize its position. Maps can be loaded into the robot, but often the robot must create a map by itself. Maps are represented using either a grid showing which cells are occupied and which are free or a continuous map containing the coordinates of the obstacles. For a grid map the frontier algorithm enables the robot to explore its environment in order to determine the probability that each cell is occupied or free. A robot can use knowledge of its environment, for example, that it is in a building with rectangular rooms, to facilitate building a map. Algorithms for simultaneous localization and mapping enable a robot to perform these two tasks together, using data from localization and its sensor to extend the map and data from the map to achieve accurate localization.

Given a map and a target position, a robot must perform path planning in order to determine the best route from its current position to the target position. Three algorithms for path planning are presented: Dijkstra’s shortest path algorithm for a grid of cells, an algorithm for continuous maps and the A algorithm, an improvement of Dijkstra’s algorithm that uses heuristic functions. The chapter concludes with a description of the integration of path planning with obstacle avoidance.