2P1-B6 自律移動ロボットの知的行動獲得に関する研究 : シンボルグラウンディングへのアプローチ(28. 創発ロボティクス)

One approach to the design of intelligent autonomous robots is through evolutionary computation. In this approach, the robot’s behavior is evolved through a process of simulated evolution, applying the Darwinian principles of survival-of-the-fittest and inheritance-with-variation to the development of the robot’s control programs. In previous studies, we illustrated this approach on problems of learning individual behaviors for autonomous mobile robots. Our previous work has focused on tasks which were reasonably complex, but which required only a single behavior. In order to scale this approach to more realistic scenarios, we now consider methods for evolving complex sets of tasks. Our approach has been to extend the basic evolutionary learning method to encompasses co-evolution, that is, the simultaneously evolution of multiple behaviors. This paper addresses alternative designs within this basic paradigm. Specifically, we focus on dependencies among the learning agents, that is, what a given learning agent needs to know about other agents in the system. By using domain knowledge, it is possible to reduce or eliminate interactions among the agents, thereby reducing the effort required to co-evolve these agents as well as reducing the impediments to learning caused by these interactions.

Expounding on the results of the author's work with the US Army Research Office, DARPA, the Office of Naval Research, and various defense industry contractors, Governing Lethal Behavior in Autonomous Robots explores how to produce an "artificial conscience" in a new class of robots, humane-oids, which are robots that can potentially perform more et

This paper describes the design, implementation and comparison of two MIMO fuzzy logic controllers for corridor following behaviour of mobile robot in indoor environments. Both the controllers take input from two ultrasonic sensors mounted on front-left and front-right sides of the robot and generate commands for left and right motors. The controllers are designed using MATLAB® and their rule bases are optimized for best possible results. The controllers are then implemented in real time using a low cost, readily available AT89C52 microcontroller. The comparison of controllers is carried out on the basis of selecting membership functions and number of rules in the rule base. Experimental results have shown the validity of both the controllers for corridor following behaviour of mobile robot.

This paper describes a robot programming course for 3rd-grade undergraduate students at the department of mechano-informatics, the university of Tokyo. This course includes many issues for building a mobile robot's behavior programming: Programming in embedded Linux environment, image processing, inverse kinematics for an arm and an omnidirectional mobile robot, etc. These issues are integrated as mobile robot's behaviors where robots explore the environment using vision and other sensors and act on the environment using omnidirectional wheels and an arm.

In this paper, we implement a computational model of a neuromodulatory system in an autonomous robot. The model is based on a set of anatomical and physiological properties of the mammalian dopamine system, one of several diffuse ascending systems of the brain. The output of this system acts as a value signal, which modulates widely distributed synaptic changes in sensory and motor areas. During reward conditioning, the model learns to generate tonic and phasic responses, which are consistent with potential roles as reward predictions and prediction errors. Different sets of neural units generate precisely timed signals that exert positive effects (predictive) and negative effects (if a predicted reward is omitted) on neuroplasticity. We test the learning and behavior of the robot in different environmental contexts, and observe changes in the development of neural connections within the neuromodulatory system that depend on the robot's interaction with the environment. Simulation of a computational model responsive to rewarding stimuli leads to the emergence of conditioned appetitive behaviors. These studies represent a step towards investigating the behavior of autonomous robots controlled by biologically based neuromodulatory systems.

For educational purposes there is a need to teach electrical and computer engineering students the basics of the design of state machines using programmable logic devices, and for students interested in mobile robots to teach them the basics of mobile robots' behaviors. At the same time one of the topics of interest in the mobile robot's community is how to generate their new behaviors, using state machines, automatically. This paper discusses how to create mobile robots' behaviors using genetic algorithms (GAs), implementing them in programmable logic devices (FPGAs) and how these behaviors can be programmed by students in a mobile robotics course. The behaviors are encoded as algorithm state machines and using feed-forward artificial neural networks (ANN). In the state-machine approach, each individual's chromosome represents, given a set of inputs coming from the sensors and the current state, both the next state and the outputs that control the robot's movements. In the ANN approach, whose weights are found also with GAs, a pipeline architecture was built to perform it, each pipeline executes a layer of the ANN, thus once the pipeline is full the execution speed of the ANN is one sensors' clock cycle. We evaluate the behaviors generated by the GA according to a fitness function that grades their performance for avoiding obstacles. The inputs to such a robot are infrared sensors to detect obstacles, the outputs are the velocities of its wheels. Our objectives are first, to prove that GAs are a good option as a method for finding behaviors for mobile robots' navigation, and second, that these behaviors can be implemented in an efficient way in FPGAs. We tested both behaviors in a small mobile robot, that is build in an electrical engineering course that teaches how to build mobile robots.

Humans not only perceive the environment but also comprehend it while moving through a scene. In contrast, most currently deployed autonomous mobile robots rely solely on perception for obstacle avoidance and route planning, completely neglecting scene understanding. This work introduces a unified framework that integrates real-time scene classification with obstacle avoidance to enable adaptive behavior in mobile robots, adjusting speed and direction based on the scene. We also present a novel indoor scene classification dataset to improve scene awareness. The effectiveness of the framework is validated through deployment on a mobile robot, demonstrating promising results and potential scalability to self-driving cars and autonomous vehicles. The demo video is available at: https://youtu.be/ija7tZ0ZypE

Interactive robots not only need to react in predefined or deterministic scenarios but also learn and adapt in real-time, mirroring cognitive flexibility akin to human intelligence. Achieving this autonomy entails developing cognitive architectures that integrate reactive, deliberative and emergent capabilities. Thus, this paper presents MERLIN2, a hybrid cognitive architecture to generate, control, plan, and monitor behaviors in autonomous robots. This architecture combines reactive, deliberative, and emergent components, aiming to enhance adaptability in dynamic environments and make intelligent real-time decisions, thereby improving autonomy and performance. MERLIN2 comprises a deliberative system, based on a knowledge base and a symbolic planner; and a behavioral system composed of reactive components and several emergent components. It addresses core cognitive aspects like action selection, perception, memory, learning, reasoning, and explainability. MERLIN2 is evaluated in a simulated world and in the real world Carry My Luggage task from the RoboCup@Home. Therefore, the experimentation presented in this article showcases the architecture as a valid solution for autonomous robots.

The behavior of an autonomous mobile robot must be executed based on the sensor information. The sensor-based behavior cannot be represented by a simple sequence of motions. Action mode representation and ROBOL/0 language are a powerful method to represent such sensor-based behavior. Autonomous navigation with the obstacle avoidance function is designed and represented using action modes. To show the effectiveness of the method, it is programmed in ROBOL/0 and executed on a real robot. >

Autonomous robots need to learn intelligent behaviour by interacting with the real world and adapting to changing environments. In order to study the adaptive behaviour of intelligent systems, an autonomous humanoid robot has been developed which understands goals, senses its environment in context with the goal and operates independently by means of experience gathering and a self-generating will.

We review current and emerging knowledge-informed and brain-inspired cognitive systems for realizing adversarial defenses, eXplainable Artificial Intelligence (XAI), and zero-shot or few-shot learning. Data-driven machine learning models have achieved remarkable performance and demonstrated capabilities surpassing humans in many applications. Yet, their inability to exploit domain knowledge leads to serious performance limitations in practical applications. In particular, deep learning systems are exposed to adversarial attacks, which can trick them into making glaringly incorrect decisions. Moreover, complex data-driven models typically lack interpretability or explainability, i.e., their decisions cannot be understood by human subjects. Furthermore, models are usually trained on standard datasets with a closed-world assumption. Hence, they struggle to generalize to unseen cases during inference in practical open-world environments, thus, raising the zero- or few-shot generalization problem. Although many conventional solutions exist, explicit domain knowledge, brain-inspired neural networks and cognitive architectures offer powerful new dimensions towards alleviating these problems. Prior knowledge is represented in appropriate forms like mathematical relations, logic rules, knowledge graphs, and large language models (LLMs). and incorporated in deep learning frameworks to improve performance. Brain-inspired cognition methods use computational models that mimic the human brain to enhance intelligent behavior in artificial agents and autonomous robots. Ultimately, these models achieve better explainability, higher adversarial robustness and data-efficient learning, and can, in turn, provide insights for cognitive science and neuroscience—that is, to deepen human understanding on how the brain works in general, and how it handles these problems.

In this paper we motivate how model-based approaches can be used in all phases of the lifetime of an autonomous robot to improve the dependability. In particular we present preliminary results of a model-based runtime diagnosis and repair approach that is applied during operation of an industrial transport robot.

In this paper programming tools for developing a sensor based behavior of a mobile robot are discussed.A method to describe a program is important base for designing programming tools. The authors have introduced an action mode representation for analyzing a mobile robot's behavior and a programming language ROBOL/0 to describe programs.The authors designed the programming tools based on the action mode representation and ROBOL/0, which is considered with (1) Action mode representation converter, (2) Simulator, (3) Action mode transition monitor, and (4) Action mode transition tracer.These tools have been implemented on UNIX workstation and used for programming of authours' mobile robot. And, it was verified that they are useful as a programming environment of autonomous behavior.

A new approach is presented which integrates evolutionary computation and real world devices such as mobile robots and an omnidirectional vision system. Starting with an evolutionary composition system named JaVOX, a hybrid environment named AURAL evolved. In the AURAL, the behavior of mobile robots in an arena is applied as a compositional strategy. It uses trajectories produced by mobile robots to modify the fitness function of a real time sonic composition. The model is described, its evolutionary design and how the interaction between the real world devices was implemented. This research is oriented towards the study of automatic and semi-automatic processes of artistic production in the sound domain.KeywordsAlgorithmic compositionevolutionary computationrobotics sonification

The problem of correctly evaluating noisy and incorrect data for the interpretation of ultrasonic sensor signals is addressed. Neural networks, with their inherent characteristics of adaptivity and high fault and noise tolerance, are well suited for such tasks. A backpropagation algorithm is described for the control of the tracking behavior of an autonomous mobile robot. Input data are provided by three ultrasonic sensors mounted on the front of the vehicle. For more flexibility the behavior and learning capability of the tracking algorithm have been improved using different networks.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>