Integrating neural networks and knowledge-based systems for robotic control (original) (raw)
Related papers
Integrating Knowledge-Based System and Neural Network Techniques for Robotic Skill Acquisition
This paper describes an approach to robotic control that is patterned after models of human skill acquisition. The intent is to develop robots capable of learning how to accomplish complex tasks using designer-supplied instructions and self induced practice. A simulation is presented in which a rule-based system supervises the training of a neural network and controls the operation of the system during the learning process. Simulation results show the interaction between rule-based and network-based system components during various phases of training and supervision.
1995
System modelling methods and feed-forward neural networks are two important tools for analysing and controlling a physical system. The system modelling methods give a clear description of the relationship between inputs and outputs of the system, based upon prior knowledge of the system, whereas neural networks can determine this relationship whilst learning examples of particular input-output pairs of the system. Because only one of these two methods is used to describe physical systems in many applications, the advantages of these two methods have not been fully utilised. In practice, every real system can be divided into two parts. We can define the describable part of a system as the part which can be described by an explicit physical-mathematical model. The system's modelling error is referred to as the unknown part. In order to find input-output relationships of systems more accurately and more efficiently, modelling methods and neural networks should be used in a hybrid f...
A neural network-based approach to robot motion control
2008
The joint controllers used in robots like the Sony Aibo are designed for the task of moving the joints of the robot to a given position. However, they are not well suited to the problem of making a robot move through a desired trajectory at speeds close to the physical capabilities of the robot, and in many cases, they cannot be bypassed easily. In this paper, we propose an approach that models both the robot's joints and its built-in controllers as a single system that is in turn controlled by a neural network.
Neural network-based modeling of robot action effects in conceptual state space of real world
Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems. IROS '96, 1996
This paper gives the concept of an autonomous robotic agent that is capable of showing both machine learning and reactive behavior. The first methodology is used to collect information about the environment and to plan robot actions based on this information while the robot is performing tasks. Processing and storing information obtained during several task executions is called lifelong learning. Reactive behavior, the second desirable feature of an autonomous robot, is needed to execute the actions in a dynamically changing environment.
A Relational Representation for Generalized Knowledge in Robotic Tasks
In this paper, a novel representation is proposed in which experience is summarized by a wealth of control and perception primitives that can be mined to learn combinations of which features are most predictive of task success. Exploiting the inherent relational structure of these primitives and the dependencies between them presents a powerful and widely-applicable new approach in the robotics community. These dependencies are represented as links in a relational dependency network (RDN), and capture information about how a robot's actions and observations affect each other when used together in the full context of the task. For example, a RDN trained as an expert to "pick up" things will represent the best way to reach to an object, knowing that it plans on grasping that object later. Such experts provide information which might not be obvious to a programmer ahead of time, and can be consulted to allow the robot to achieve higher levels of task performance. Furthermore, it seems possible that new, more complex RDNs could be trained by learning the dependencies between existing RDNs. As a result, this paper proposes a hierarchical way of organizing complex behaviors in a principled way.
Novel automated interactive reinforcement learning framework with a constraint-based supervisor for procedural tasks, 2024
Learning to perform procedural motion or manipulation tasks in unstructured or uncertain environments poses significant challenges for intelligent agents. Although reinforcement learning algorithms have demonstrated positive results on simple tasks, the hard-to-engineer reward functions and the impractical amount of trialand-error iterations these agents require in long-experience streams still present challenges for deployment in industrially relevant environments. In this regard, interactive reinforcement learning has emerged as a promising approach to mitigate these limitations, whereby a human supervisor provides evaluative or corrective feedback to the learning agent during training. However, the requirement of a human-in-the-loop approach throughout the learning process can be impractical for tasks that span several hours. This study aims to overcome this limitation by automating the learning process and substituting human feedback with an artificial supervisor grounded in constraint-based modeling techniques. In contrast to the logical constraints commonly used for conventional reinforcement learning, constraint-based modeling techniques offer enhanced adaptability in terms of conceptualizing and modeling the human knowledge of a task. This modeling capability allows an automated supervisor to acquire a closer approximation to human reasoning by dividing complex tasks into more manageable components and identifying the associated subtask and contextual cues in which the agent is involved. The supervisor then adjusts the evaluative and corrective feedback to suit the specific subtask under consideration. The framework was assessed using three actor-critic agents in a human-robot interaction environment, demonstrating a sample efficiency improvement of 50% and success rates of ≥95% in simulation and 90% in real-world implementation.
Design of a Neural Interface Based System for Control of Robotic Devices
The paper describes the design of a Neural Interface Based (NIS) system for control of external robotic devices. The system is being implemented using the principles of component-based reuse and combines modules for data acquisition, data processing, training, classification, direct and the NIS-based control as well as evaluation and graphical representation of results. The system uses the OCZ Neural Impulse Actuator to acquire the data for control of Arduino 4WD and Lynxmotion 5LA Robotic Arm devices. The paper describes the implementation of the system's components as well as presents the results of experiments.
COULD KNOWLEDGE-BASED NEURAL LEARNING BE USEFUL IN DEVELOPMENTAL ROBOTICS? THE CASE OF KBCC
International Journal of Humanoid Robotics, 2007
The new field of developmental robotics faces the formidable challenge of implementing effective learning mechanisms in complex, dynamic environments. We make a case that knowledge-based learning algorithms might help to meet this challenge. A constructive neural learning algorithm, knowledge-based cascade-correlation (KBCC), autonomously recruits previously-learned networks in addition to the single hidden units recruited by ordinary cascade-correlation. This enables learning by analogy when adequate prior knowledge is available, learning by induction from examples when there is no relevant prior knowledge, and various combinations of analogy and induction. A review of experiments with KBCC indicates that recruitment of relevant existing knowledge typically speeds learning and sometimes enables learning of otherwise impossible problems. Some additional domains of interest to developmental robotics are identified in which knowledge-based learning seems essential. The characteristics of KBCC in relation to other knowledge-based neural learners and analogical reasoning are summarized as is the neurological basis for learning from knowledge. Current limitations of this approach and directions for future work are discussed.
Hierarchical intelligent control for robotic motion
IEEE Transactions on Neural Networks, 1994
This paper presents a new scheme for intelligent control of robotic manipulators. This scheme is a hierarchically integrated approach to neuromorphic and symbolic control of robotic manipulators. This includes an applied neural network for servo control and knowledge-based approximation. The neural network in the servo control level is based on a numerical manipulation, while the knowledge based part is symbolic manipulation. The knowledge base part develops control strategies symbolically for the servo level. The neural network compensates for vagueness in the control strategies, nonlinearities of the system and uncertainties in its environment using neuromorphic control.