[...]... concentrate on the concept of reactive control to generate the behavior of four- and six-legged walking machines In particular, we shall present a behavior controller based on a modular neural structure with an artificial neural network using discrete-time dynamics It consists of two main modules: neural preprocessing and neural control1 (Fig 1.10) The function of this kind of a neural controller is easier... Indeed, most of this book is aimed at explaining how to: • Use a modular neural structure where the neural control unit can be coupled with the different neural preprocessing units to form the desired behavior controls The neural structures are simple to understand and can be applied to control different types of walking machines • Minimize the complexity of the neural preprocessing and control unit... the neural preprocessing and control can be formed using a small neural structure In order to physically build four- and six-legged walking machines for testing and demonstrating the capability of the behavior controllers, the morphologies of walking animals are used as inspiration for the design The basic locomotion control of the walking machines is also created by determining the principle of animal... different reactive behavior controls can be created by coupling the neural control module with different neural preprocessing modules Because the functionality of the modules is well understood, the reactive behavior controller of a less complex agent2 (four-legged walking machine) can be applied also to a more complex agent (six-legged walking machine), and vice versa A part of 1 2 Here, neural preprocessing. .. of the combination between the different neural preprocessing units and the neural control unit It ends up with the detail of behavior fusion control that combines all created reactive behaviors and leads to versatile artificial perception–action systems Chapter 6 shows the detailed results of the neural preprocessing tested with the simulated and real sensory signals It also shows the capabilities of. .. determines the boundary conditions of an environment in which it can operate successfully The design of the neural control depends on the morphology of the agent, i.e., the type and position of the sensors and the configurations of the actuators Choosing too simple a design, the behavior of the body may be of limited interest and it may obstruct the need for an effective neural control for a complex system... consists of two methods: a look-up table, which manages sensory input by referring to their predefined priorities, and a time scheduling method, which switches behavioral modes 12 1 Introduction Chapter 5, which is the main contribution of this book, introduces the neural preprocessing of sensory signals and neural control for the locomotion of walking machines It also presents different behavior controls... fields of study, e.g., the investigation of the robotic behavior control and the understanding of how a biological system works It is also the basis for the creation of a so-called Autonomous Intelligent System, which is an active area of research and a highly challenging field Thus, the work described here continues in this tradition with the extension of the use of biologically inspired walking machines. .. 68 5.1.2 Preprocessing of a Tactile Signal 82 5.1.3 Preprocessing of Antenna-like Sensor Data 87 5.2 Neural Control of Walking Machines 89 5.2.1 The Neural Oscillator Network 89 5.2.2 The Velocity Regulating Network 94 5.2.3 The Modular Neural Controller 98 5.3 Behavior Control ... morphologies similar to walking animals are preferred In other words, biologically inspired walking machines are the robot platforms for the approach of this work Such machines are more attractive because they can behave somewhat like animals and they are still a challenge for locomotion control Two walking animals were observed to benefit the leg and trunk designs of four- and six-legged walking machines (physical . be- havior controls. The neural structures are simple to understand and can be applied to control different types of walking machines. • Minimize the complexity of the neural preprocessing and control. 67 5.1 NeuralPreprocessingof Sensory Signals 67 5.1.1 AuditorySignalProcessing 68 5.1.2 Preprocessingofa TactileSignal 82 5.1.3 Preprocessing of Antenna-like Sensor Data 87 5.2 Neural Control of Walking. Walsh Bonnie Webber Poramate Manoonpong Neural Preprocessing and Control of Reactive Walking Machines Towards Versatile Artificial Perception–Action Systems With 142 Figures and 3 Tables 123 Author: Poramate