Language games for autonomous robots

Integration and grounding are key AI challenges for human-robot dialogue. The author and his team are tackling these issues using language games and have experimented with them on progressively more complex platforms. A language game is a sequence of verbal interactions between two agents situated in a specific environment. Language games both integrate the various activities required for dialogue and ground unknown words or phrases in a specific context, which helps constrain possible meanings.     

Adaptive navigation for autonomous robots

In many robotic exploration missions, robots have to learn specific policies that allow them to: (i) select high level goals (e.g., identify specific destinations), (ii) navigate (reach those destinations), (iii) and adapt to their environment (e.g., modify their behavior based on changing environmental conditions). Furthermore, those policies must be robust to signal noise or unexpected situations, scalable to more complex environments, and account for the physical limitations of the robots (e.g., limited battery power and computational power).In this paper we evaluate reactive and learning navigation algorithms for exploration robots that must avoid obstacles and reach specific destinations in limited time and with limited observations. Our results show that neuro-evolutionary algorithms with well-designed evaluation functions can produce up to 50% better performance than reactive algorithms in complex domains where the robot’s goals are to select paths that lead to seek specific destinations while avoiding obstacles, particularly when facing significant sensor and actuator signal noise.     

Developmental learning for autonomous robots

Developmental robotics is concerned with the design of algorithms that promote robot adaptation and learning through qualitative growth of behaviour and increasing levels of competence.This paper uses ideas and inspiration from early infant psychology (up to three months of age) to examine how robot systems could discover the structure of their local sensory-motor spaces and learn how to coordinate these for the control of action.An experimental learning model is described and results from robotic experiments using the model are presented and discussed.     

Designing autonomous robots

Autonomous robots are complex systems that require the interaction or cooperation of numerous heterogeneous software components. Nowadays, robots are getting closer to humans and as such are becoming critical systems that must meet safety properties including logical, temporal, and real-time constraints.     

Hybrid autonomous control for multi mobile robots

Reinforcement learning can be an adaptive and flexible control method for autonomous system. It does not need a priori knowledge; behaviors to accomplish given tasks are obtained automatically by repeating trial and error. However, with increasing complexity of the system, the learning costs are increased exponentially. Thus, application to complex systems, like a many redundant d.o.f. robot and multi-agent system, is very difficult. In the previous works in this field, applications were restricted to simple robots and small multi-agent systems, and because of restricted functions of the simple systems that have less redundancy, effectiveness of reinforcement learning is restricted. In our previous works, we had taken these problems into consideration and had proposed new reinforcement learning algorithm, 'Q-learning with dynamic structuring of exploration space based on GA (QDSEGA)'. Effectiveness of QDSEGA for redundant robots has been demonstrated using a 12-legged robot and a 50-link manipulator. However, previous works on QDSEGA were restricted to redundant robots and it was impossible to apply it to multi mobile robots. In this paper, we extend our previous work on QDSEGA by combining a rule-based distributed control and propose a hybrid autonomous control method for multi mobile robots. To demonstrate the effectiveness of the proposed method, simulations of a transportation task by 10 mobile robots are carried out. As a result, effective behaviors have been obtained.     

Energetically autonomous robots: Food for thought

This paper reports on the robot EcoBot-II, which is designed to power itself solely by converting unrefined insect biomass into useful energy using on-board microbial fuel cells with oxygen cathodes. In bench experiments different ‘fuels’ (sugar, fruit and dead flies) were explored in the microbial fuel cell system and their efficiency of conversion to electricity is compared with the maximum available energy calculated from bomb calorimetry trials. In endurance tests EcoBot-II was able to run for 12 days while carrying out phototaxis, temperature sensing and radio transmission of sensed data approximately every 14 min.     

Enhanced continuous valued Q-learning for real autonomous robots

A parallel-jaw gripper is a very useful tool for robot manipulation tasks due to its simple mechanism and control. This fact limits the range of successful grasps it can undergo, and also makes it unfeasible under uncertainties. Thus, it is desirable to improve its dexterity and manipulability. In this paper, we propose a new design of a two-fingered parallel gripper that utilizes rolling at the contacts for object repositioning and reorientation, aimed at effective firm grasps. We name it the scrollic gripper, an acronym for synchronously closing with rolling constraints. At first, the background to utilize the rolling constraints is described. Then, grasping and manipulation of the gripper are discussed. In grasp acquisition, we propose a quality function for evaluating grasp stability. The sophisticated hardware and functioning for the scrollic gripper consist, basically, on implementation of an additional degree-of-freedom to the conventional parallel-jaw gripper, leading to grasp acquisition and secure grasping.     

Portable autonomous walk calibration for 4-legged robots

In the present paper we describe an efficient and portable optimization method for calibrating the walk parameters of a quadruped robot, and its contribution for the robot control and localization. The locomotion of a legged robot presents not only the problem of maximizing the speed, but also the problem of obtaining a precise speed response, and achieving an acceptable odometry information. In this study we use a simulated annealing algorithm for calibrating different parametric sets for different speed ranges, with the goal of avoiding discontinuities. The results are applied to the robot AIBO in the RoboCup domain. Moreover, we outline the relevance of calibration to the control, showing the improvement obtained in odometry and, as a consequence, in robot localization.     

Bioinspired algorithm for area surveillance using autonomous robots

Territorial surveillance plays a constantly increasing role in security. However, completely automatic surveillance using autonomous robots is hard to implement and maintain. Current methods described in literature propose systems that include direct communication of the robots or the use of a centralised system to coordinate the robots. These systems are prone to equipment failure and/or malicious attacks to the centralised system. In this paper, we propose a bioinspired algorithm that allows indirect communication between the robots that are considered minimally equipped. This is achieved by applying a parallel and distributed technique inspired by the emergent behaviour of social insects, namely ant colonies. In particular, the development of a collective memory for robots and areas covered is achieved subsequently through self-organisation of the autonomous robots to a continuous dynamic coverage of the test space. The algorithm is shown to have a robust behaviour and competitive performance. Several simulations run for various space sizes, different number of robots, different pheromone evaporation rates as well various percentages of space covered by obstacles. In all cases the efficacy of the proposed algorithm has been successfully proven when compared with other well known techniques.     

Entropy-driven on-line control of autonomous robots

A new approach to on-line path planning is derived in this paper. The planning algorithm is motivated by robot navigation and manipulation tasks in uncertain, unstructured, dynamic environments. A minimum entropy evidential classifier is used to recognize targets and obstacles in the environment. An iterative Newton scheme is then used to generate a sequence of knot points that guide the motion of the robot. The acquisition and processing of sensory data continue during the motion, thus reducing the uncertainty about the environment. The classification of targets and obstacles is updated, and the path is replanned (locally) to adapt to those changes. A graphical tool based on the concept of Julia sets is used to ensure the predictability and smoothness of the paths.     

Automatic design of fuzzy controllers for car-like autonomous robots

This paper describes the design and implementation of a fuzzy control system for a car-like autonomous vehicle. The problem addressed is the diagonal parking in a constrained space, a typical problem in motion control of nonholonomic robots. The architecture proposed for the fuzzy controller is a hierarchical scheme which combines seven modules working in series and in parallel. The rules of each module employ the adequate fuzzy operators for its task (making a decision or generating a smoothly varying control output), and they have been obtained from heuristic knowledge and numerical data (with geometric information) depending on the module requirements (some of them are constrained to provide paths of near-minimal lengths). The computer-aided design tools of the environment Xfuzzy 3.0 (developed by some of the authors) have been employed to automate the different design stages: 1) translation of heuristic knowledge into fuzzy rules; 2) extraction of fuzzy rules from numerical data and their tuning to give paths of near-minimal lengths; 3) offline verification of the control system behavior; and 4) its synthesis to be implemented in a true robot and be verified on line. Real experiments with the autonomous vehicle ROMEO 4R (designed and built at the Escuela Superior de Ingenieros, University of Seville, Seville, Spain) demonstrate the efficiency of the described controller and of the methodology followed in its design.     

Mobility characterization for autonomous mobile robots using machine learning

This paper presents a supervised learning approach to improving the autonomous mobility of wheeled robots through sensing the robot’s interaction with terrain ‘underfoot.’ Mobility characterization is cast as a hierarchical task, in which pre-immobilization detection is achieved using support vector machines in time to prevent full immobilization, and if a pre-immobilization condition is detected, the associated terrain feature affecting mobility is identified using a Hidden Markov model. These methods are implemented using a hierarchical, layered control scheme developed for the Yeti robot, a 73-kg, four-wheeled robot designed to perform autonomous medium-range missions in polar terrain. The methodology is motivated by the difficultly of visually recognizing terrain features that impact mobility in low contrast terrain. The efficacy of the approach is evaluated using data from a suite of proprioceptive sensors. Real-time implementation shows that Yeti can consistently detect pre-immobilization conditions, stop in time to avoid unrecoverable immobilization, identify the terrain feature presenting the mobility challenge, and execute an escape sequence to retreat from the condition.     

Discretization and fitting of nominal data for autonomous robots

This paper presents methodologies to discretize nominal robot paths extracted from 3-D CAD drawings. Behind robot path discretization is the ability to have a robot adjusting the traversed paths so that the contact between robot tool and work-piece is properly maintained. In addition, a hybrid force/motion control system based on Fuzzy-PI control is proposed to adjust robot paths with external sensory feedback. All these capabilities allow to facilitate the robot programming process and to increase the robot’s autonomy.     

Theoretical foundations of planning and navigation for autonomous robots

In this article, theoretical foundations of planning and navigation systems and processes are outlined in a form applicable for autonomous robotics. the core of the theory is based upon methods developed in the team theory of decentralized stochastic control. A class of autonomous control systems is defined, and a problem of information representation is addressed for this class. A phenomenon of nesting is determined to be a key element for arranging design and control procedures for systems of intelligent control. A concept of hierarchical nested knowledge-based controller is employed in this article which enables optimum control using nested dynamic programming. an application of this concept is unfolded for a system of knowledge-based control of an autonomous mobile robot.     

Fuzzy rule-based environment-aware autonomous mobile robots for actuated touring

Intelligent Service Robotics - The involvement of computer-programmed autonomous mobile robots in real-time activities is emerging in the recent years. The actuation and interaction of the robots...     

A chaotic path planning generator for autonomous mobile robots

This work presents a chaotic path planning generator which is used in autonomous mobile robots, in order to cover a terrain. The proposed generator is based on a nonlinear circuit, which shows chaotic behavior. The bit sequence, produced by the chaotic generator, is converted to a sequence of planned positions, which satisfies the requirements for unpredictability and fast scanning of the entire terrain. The nonlinear circuit and the trajectory-planner are described thoroughly. Simulation tests confirm that with the proposed path planning generator better results can be obtained with regard to previous works.     

A membrane computing framework for self-reconfigurable robots

Self-reconfigurable robots are built by modules which can move in relationship to each other, which allows the robot to change its physical form. Finding a sequence of module moves that reconfigures the robot from the initial configuration to the goal configuration is a hard task and many control algorithms have been proposed. In this paper, we present a novel method which combines a cluster-flow locomotion based on cellular automata together with a decentralized local representation of the spatial geometry based on membrane computing ideas. This new approach has been tested with computer simulations and real-world experiments performed with modular self-reconfigurable robots and represents a new point of view with respect other control methods found in the literature.

     

Development environments for autonomous mobile robots: A survey

Robotic Development Environments (RDEs) have come to play an increasingly important role in robotics research in general, and for the development of architectures for mobile robots in particular. Yet, no systematic evaluation of available RDEs has been performed; establishing a comprehensive list of evaluation criteria targeted at robotics applications is desirable that can subsequently be used to compare their strengths and weaknesses. Moreover, there are no practical evaluations of the usability and impact of a large selection of RDEs that provides researchers with the information necessary to select an RDE most suited to their needs, nor identifies trends in RDE research that suggest directions for future RDE development. This survey addresses the above by selecting and describing nine open source, freely available RDEs for mobile robots, evaluating and comparing them from various points of view. First, based on previous work concerning agent systems, a conceptual framework of four broad categories is established, encompassing the characteristics and capabilities that an RDE supports. Then, a practical evaluation of RDE  usability in designing, implementing, and executing robot architectures is presented. Finally, the impact of specific RDEs on the field of robotics is addressed by providing a list of published applications and research projects that give concrete examples of areas in which systems have been used. The comprehensive evaluation and comparison of the nine RDEs concludes with suggestions of how to use the results of this survey and a brief discussion of future trends in RDE design.