Abstract
Neural networks have been used in a number of robotic applications (Das & Kar, 2006; Fierro & Lewis, 1998), including both manipulators and mobile robots. A typical approach is to use neural networks for nonlinear system modelling, including for instance the learning of forward and inverse models of a plant, noise cancellation, and other forms of nonlinear control (Fierro & Lewis, 1998). An alternative approach is to solve a particular problem by designing a specialized neural network architecture and/or learning rule (Sutton & Barto, 1981). It is clear that biological brains, though exhibiting a certain degree of homogeneity, rely on many specialized circuits designed to solve particular problems. We are interested in understanding how animals are able to solve complex problems such as learning to navigate in an unknown environment, with the aim of applying what is learned of biology to the control of robots (Chang & Gaudiano, 1998; Martínez-Marín, 2007; Montes-González, Santos-Reyes & Ríos- Figueroa, 2006). In particular, this article presents a neural architecture that makes possible the integration of a kinematical adaptive neuro-controller for trajectory tracking and an obstacle avoidance adaptive neuro-controller for nonholonomic mobile robots. The kinematical adaptive neuro-controller is a real-time, unsupervised neural network that learns to control a nonholonomic mobile robot in a nonstationary environment, which is termed Self-Organization Direction Mapping Network (SODMN), and combines associative learning and Vector Associative Map (VAM) learning to generate transformations between spatial and velocity coordinates (García-Córdova, Guerrero-González & García-Marín, 2007). The transformations are learned in an unsupervised training phase, during which the robot moves as a result of randomly selected wheel velocities. The obstacle avoidance adaptive neurocontroller is a neural network that learns to control avoidance behaviours in a mobile robot based on a form of animal learning known as operant conditioning. Learning, which requires no supervision, takes place as the robot moves around a cluttered environment with obstacles. The neural network requires no knowledge of the geometry of the robot or of the quality, number, or configuration of the robot’s sensors. The efficacy of the proposed neural architecture is tested experimentally by a differentially driven mobile robot.
TopBackground
Several heuristic approaches based on neural networks (NNs) have been proposed for identification and adaptive control of nonlinear dynamic systems (Fierro & Lewis, 1998; Pardo-Ayala & Angulo-Bahón, 2007).
In wheeled mobile robots (WMR), the trajectory-tracking problem with exponential convergence has been solved theoretically using time-varying state feedback based on the backstepping technique in (Ping & Nijmeijer, 1997; Das & Kar, 2006). Dynamic feedback linearization has been used for trajectory tracking and posture stabilization of mobile robot systems in chained form (Oriolo, Luca & Vendittelli, 2002).
The study of autonomous behaviour has become an active research area in the field of robotics. Even the simplest organisms are capable of behavioural feats unimaginable for the most sophisticated machines. When an animal has to operate in an unknown environment it must somehow learn to predict the consequences of its own actions. Biological organisms are a clear example that this short of learning is possible in spite of what, from an engineering standpoint, seem to be insurmountable difficulties: noisy sensors, unknown kinematics and dynamics, nostationary statistics, and so on. A related form of learning is known as operant conditioning (Grossberg, 1971). Chang and Gaudiano (1998) introduce a neural network for obstacle avoidance that is based on a model of classical and operant conditioning.
Key Terms in this Chapter
Artificial Neural Network: A network of many simple processors (“units” or “neurons”) that imitates a biological neural network. The units are connected by unidirectional communication channels, which carry numeric data. Neural networks can be trained to find nonlinear relationships in data; and are used in applications such as robotics, speech recognition, and signal processing or medical diagnosis.
Operant Conditioning: The term “Operant” refers to how an organism operates on the environment, and hence, operant conditioning comes from how we respond to what is presented to us in our environment. Then the operant conditioning is a form of associative learning through which an animal learns about the consequences of its behaviour.
Unconditioned Stimulus (UCS): Which is one that unconditionally, naturally, and automatically triggers an innate, often reflexive, response in the presence of significant stimulus. For example, when you smell one of your favourite foods, you may immediately feel very hungry. In this example, the smell of the food is the unconditioned stimulus.
Conditioned Stimulus (CS): It is a previously neutral stimulus that, after becoming associated with the unconditioned stimulus, eventually comes to trigger a conditioned response. The neutral stimulus could be any event that does not result in an overt behavioral response from the organism under investigation.
Unconditioned Response (UR): It is the unlearned response that occurs naturally in response to the unconditioned stimulus.
Classical Conditioning: It is a form of associative learning that was first demonstrated by Ivan Pavlov. The typical procedure for inducing classical conditioning involves a type of learning in which a stimulus acquires the capacity to evoke a response that was originally evoked by another stimulus.
Conditioned Response (CR): If the conditioned stimulus and the unconditioned stimulus are repeatedly paired, eventually the two stimuli become associated and the organism begins to produce a behavioral response to the conditioned stimulus. Then, the conditioned response is the learned response to the previously neutral stimulus.