Distributing a neural network among the robots of a swarm

September 3, 2018 15:44 ,

Michael Otte, An emergent group mind across a swarm of robots: Collective cognition and distributed sensing via a shared wireless neural network, The International Journal of Robotics Research, DOI: 10.1177/0278364918779704.

We pose the “trained-at-runtime heterogeneous swarm response problem,” in which a swarm of robots must do the following three things: (1) Learn to differentiate between multiple classes of environmental feature patterns (where the feature patterns are distributively sensed across all robots in the swarm). (2) Perform the particular collective behavior that is the appropriate response to the feature pattern that the swarm recognizes in the environment at runtime (where a collective behavior is defined by a mapping of robot actions to robots). (3) The data required for both (1) and (2) is uploaded to the swarm after it has been deployed, i.e., also at runtime (the data required for (1) is the specific environmental feature patterns that the swarm should learn to differentiate between, and the data required for (2) is the mapping from feature classes to swarm behaviors). To solve this problem, we propose a new form of emergent distributed neural network that we call an “artificial group mind.” The group mind transforms a robotic swarm into a single meta-computer that can be programmed at runtime. In particular, the swarm-spanning artificial neural network emerges as each robot maintains a slice of neurons and forms wireless neural connections between its neurons and those on nearby robots. The nearby robots are discovered at runtime. Experiments on real swarms containing up to 316 robots demonstrate that the group mind enables collective decision-making based on distributed sensor data, and solves the trained-at-runtime heterogeneous swarm response problem. The group mind is a new tool that can be used to create more complex emergent swarm behaviors. The group mind also enables swarm behaviors to be a function of global patterns observed across the environment—where the patterns are orders of magnitude larger than the robots themselves.

Posted in: Robot task planning , Tagged: ,

Convergence in reference tracking by a nonlinear system, with a known model, remotely controlled through WiFi

July 27, 2018 07:52 ,

Ali Parsa, Alireza Farhadi, Measurement and control of nonlinear dynamic systems over the internet (IoT): Applications in remote control of autonomous vehicles, Automatica, Volume 95, 2018, Pages 93-103 DOI: 10.1016/j.automatica.2018.05.016.

This paper presents a new technique for almost sure asymptotic state tracking, stability and reference tracking of nonlinear dynamic systems by remote controller over the packet erasure channel, which is an abstract model for transmission via WiFi and the Internet. By implementing a suitable linearization method, a proper encoder and decoder are presented for tracking the state trajectory of nonlinear systems at the end of communication link when the measurements are sent through the packet erasure channel. Then, a controller for reference tracking of the system is designed. In the proposed technique linearization is applied when the error between the states and an estimate of these states at the decoder increases. It is shown that the proposed technique results in almost sure asymptotic reference tracking (and hence stability) over the packet erasure channel. The satisfactory performance of the proposed state trajectory and reference tracking technique is illustrated by computer simulations by applying this technique on the unicycle model, which represents the dynamic of autonomous vehicles.

Posted in: Networked telerobots

A review on mobile robot navigation

July 25, 2018 08:15 ,

Tzafestas, S.G. , Mobile Robot Control and Navigation: A Global Overview,J Intell Robot Syst (2018) 91: 35 DOI: 10.1007/s10846-018-0805-9.

The aim of this paper is to provide a global overview of mobile robot control and navigation methodologies developed over the last decades. Mobile robots have been a substantial contributor to the welfare of modern society over the years, including the industrial, service, medical, and socialization sectors. The paper starts with a list of books on autonomous mobile robots and an overview of survey papers that cover a wide range of decision, control and navigation areas. The organization of the material follows the structure of the author’s recent book on mobile robot control. Thus, the following aspects of wheeled mobile robots are considered: kinematic modeling, dynamic modeling, conventional control, affine model-based control, invariant manifold-based control, model reference adaptive control, sliding-mode control, fuzzy and neural control, vision-based control, path and motion planning, localization and mapping, and control and software architectures.

Posted in: Robot motion planning , Tagged: ,

A formal definition of autonomy and of its degrees

July 25, 2018 08:12 ,

Antsaklis, P.J. & Rahnama, A. , Control and Machine Intelligence for System Autonomy, Journal of Intelligent & Robotic Systems
July 2018, Volume 91, Issue 1, pp 23–34 DOI: 10.1007/s10846-018-0832-6.

Autonomous systems evolve from control systems by adding functionalities that increase the level of system autonomy. It is very important to the research in the field that autonomy be well defined and so in the present paper a precise, useful definition of autonomy is introduced and discussed. Autonomy is defined as the ability of the system to attain a set of goals under a set of uncertainties. This leads to the notion of degrees or levels of autonomy. The Quest for Autonomy in engineered systems throughout the centuries is noted, connections to research work of 30 years ago are made and a hierarchical functional architecture for autonomous systems together with needed functionalities are outlined. Adaptation and Learning, which are among the most important functions in achieving high levels of autonomy are then highlighted and recent research contributions are briefly discussed.

Posted in: Control Engineering, Systems Engineering , Tagged:

A parallel implementation of a new probabilistic filter for occupancy grid maps that deals with non-static environments

July 25, 2018 08:10 ,

Dominik Nuss, Stephan Reuter, Markus Thom, …, A random finite set approach for dynamic occupancy grid maps with real-time application, The International Journal of Robotics Research DOI: 10.1177/0278364918775523.

Grid mapping is a well-established approach for environment perception in robotic and automotive applications. Early work suggests estimating the occupancy state of each grid cell in a robot’s environment using a Bayesian filter to recursively combine new measurements with the current posterior state estimate of each grid cell. This filter is often referred to as binary Bayes filter. A basic assumption of classical occupancy grid maps is a stationary environment. Recent publications describe bottom-up approaches using particles to represent the dynamic state of a grid cell and outline prediction-update recursions in a heuristic manner. This paper defines the state of multiple grid cells as a random finite set, which allows to model the environment as a stochastic, dynamic system with multiple obstacles, observed by a stochastic measurement system. It motivates an original filter called the probability hypothesis density / multi-instance Bernoulli (PHD/MIB) filter in a top-down manner. The paper presents a real-time application serving as a fusion layer for laser and radar sensor data and describes in detail a highly efficient parallel particle filter implementation. A quantitative evaluation shows that parameters of the stochastic process model affect the filter results as theoretically expected and that appropriate process and observation models provide consistent state estimation results.

Posted in: Mobile robot mapping , Tagged: ,

A survey on the concept of Entropy as a measure of the intelligence and autonomy of a system, modeled hierarchically

July 25, 2018 08:07 ,

Valavanis, K.P., The Entropy Based Approach to Modeling and Evaluating Autonomy and Intelligence of Robotic Systems, J Intell Robot Syst (2018) 91: 7 DOI: 10.1007/s10846-018-0905-6.

This review paper presents the Entropy approach to modeling and performance evaluation of Intelligent Machines (IMs), which are modeled as hierarchical, multi-level structures. It provides a chronological summary of developments related to intelligent control, from its origins to current advances. It discusses fundamentals of the concept of Entropy as a measure of uncertainty and as a control function, which may be used to control, evaluate and improve through adaptation and learning performance of engineering systems. It describes a multi-level, hierarchical, architecture that is used to model such systems, and it defines autonomy and machine intelligence for engineering systems, with the aim to set foundations necessary to tackle related challenges. The modeling philosophy for the systems under consideration follows the mathematically proven principle of Increasing Precision with Decreasing Intelligence (IPDI). Entropy is also used in the context of N-Dimensional Information Theory to model the flow of information throughout such systems and contributes to quantitatively evaluate uncertainty, thus, autonomy and intelligence. It is explained how Entropy qualifies as a unique, single, measure to evaluate autonomy, intelligence and precision of task execution. The main contribution of this review paper is that it brings under one forum research findings from the 1970’s and 1980’s, and that it supports the argument that even today, given the unprecedented existing computational power, advances in Artificial Intelligence, Deep Learning and Control Theory, the same foundational framework may be followed to study large-scale, distributed Cyber Physical Systems (CPSs), including distributed intelligence and multi-agent systems, with direct applications to the SmartGrid, transportation systems and multi-robot teams, to mention but a few applications.

Posted in: Control Engineering, Systems Engineering , Tagged: , ,

High performance robotic computing (HPRC) vs. High performance computing, and its application to multirobot systems

July 17, 2018 11:43 ,

Leonardo Camargo-Forero, Pablo Royo, Xavier Prats, Towards high performance robotic computing, Robotics and Autonomous Systems, Volume 107, 2018, Pages 167-181 DOI: 10.1016/j.robot.2018.05.011.

Embedding a robot with a companion computer is becoming a common practice nowadays. Such computer is installed with an operatingsystem, often a Linux distribution. Moreover, Graphic Processing Units (GPUs) can be embedded on a robot, giving it the capacity of performing complex on-board computing tasks while executing a mission. It seems that a next logical transition, consist of deploying a cluster of computers among embedded computing cards. With this approach, a multi-robot system can be set as a High Performance Computing (HPC) cluster. The advantages of such infrastructure are many, from providing higher computing power up to setting scalable multi-robot systems. While HPC has been always seen as a speeding-up tool, we believe that HPC in the world of robotics can do much more than simply accelerating the execution of complex computing tasks. In this paper, we introduce the novel concept of High Performance Robotic Computing — HPRC, an augmentation of the ideas behind traditional HPC to fit and enhance the world of robotics. As a proof of concept, we introduce novel HPC software developed to control the motion of a set of robots using the standard parallel MPI (Message Passing Interface) library. The parallel motion software includes two operation modes: Parallel motion to specific target and swarm-like behavior. Furthermore, the HPC software is virtually scalable to control any quantity of moving robots, including Unmanned Aerial Vehicles, Unmanned Ground Vehicles, etc.

Posted in: Computer science, Robotic software , Tagged: ,

Filters with quaternions for localization

July 3, 2018 11:35 ,

Rangaprasad Arun Srivatsan, Mengyun Xu, Nicolas Zevallos, and Howie Choset, Probabilistic pose estimation using a Bingham distribution-based linear filter, The International Journal of Robotics Research DOI: 10.1177/0278364918778353.

Pose estimation is central to several robotics applications such as registration, hand–eye calibration, and simultaneous localization and mapping (SLAM). Online pose estimation methods typically use Gaussian distributions to describe the uncertainty in the pose parameters. Such a description can be inadequate when using parameters such as unit quaternions that are not unimodally distributed. A Bingham distribution can effectively model the uncertainty in unit quaternions, as it has antipodal symmetry, and is defined on a unit hypersphere. A combination of Gaussian and Bingham distributions is used to develop a truly linear filter that accurately estimates the distribution of the pose parameters. The linear filter, however, comes at the cost of state-dependent measurement uncertainty. Using results from stochastic theory, we show that the state-dependent measurement uncertainty can be evaluated exactly. To show the broad applicability of this approach, we derive linear measurement models for applications that use position, surface-normal, and pose measurements. Experiments assert that this approach is robust to initial estimation errors as well as sensor noise. Compared with state-of-the-art methods, our approach takes fewer iterations to converge onto the correct pose estimate. The efficacy of the formulation is illustrated with a number of examples on standard datasets as well as real-world experiments.

Posted in: Mobile Robot Localization

A robot architecture for humanoids able to coordinate different cognitive processes (perception, decision-making, etc.) in a hierarchical fashion

June 14, 2018 08:59 ,

J. Hwang and J. Tani, Seamless Integration and Coordination of Cognitive Skills in Humanoid Robots: A Deep Learning Approach, IEEE Transactions on Cognitive and Developmental Systems, vol. 10, no. 2, pp. 345-358 DOI: 10.1109/TCDS.2017.2714170.

This paper investigates how adequate coordination among the different cognitive processes of a humanoid robot can be developed through end-to-end learning of direct perception of visuomotor stream. We propose a deep dynamic neural network model built on a dynamic vision network, a motor generation network, and a higher-level network. The proposed model was designed to process and to integrate direct perception of dynamic visuomotor patterns in a hierarchical model characterized by different spatial and temporal constraints imposed on each level. We conducted synthetic robotic experiments in which a robot learned to read human’s intention through observing the gestures and then to generate the corresponding goal-directed actions. Results verify that the proposed model is able to learn the tutored skills and to generalize them to novel situations. The model showed synergic coordination of perception, action, and decision making, and it integrated and coordinated a set of cognitive skills including visual perception, intention reading, attention switching, working memory, action preparation, and execution in a seamless manner. Analysis reveals that coherent internal representations emerged at each level of the hierarchy. Higher-level representation reflecting actional intention developed by means of continuous integration of the lower-level visuo-proprioceptive stream.

Posted in: Developmental robotics , Tagged: , ,

An interesting model of Basal Ganglia that performs similarly to Q learning when applied to a robot

June 14, 2018 08:55 ,

Y. Zeng, G. Wang and B. Xu, A Basal Ganglia Network Centric Reinforcement Learning Model and Its Application in Unmanned Aerial Vehicle, IEEE Transactions on Cognitive and Developmental Systems, vol. 10, no. 2, pp. 290-303 DOI: 10.1109/TCDS.2017.2649564.

Reinforcement learning brings flexibility and generality for machine learning, while most of them are mathematical optimization driven approaches, and lack of cognitive and neural evidence. In order to provide a more cognitive and neural mechanisms driven foundation and validate its applicability in complex task, we develop a basal ganglia (BG) network centric reinforcement learning model. Compared to existing work on modeling BG, this paper is unique from the following perspectives: 1) the orbitofrontal cortex (OFC) is taken into consideration. OFC is critical in decision making because of its responsibility for reward representation and is critical in controlling the learning process, while most of the BG centric models do not include OFC; 2) to compensate the inaccurate memory of numeric values, precise encoding is proposed to enable working memory system remember important values during the learning process. The method combines vector convolution and the idea of storage by digit bit and is efficient for accurate value storage; and 3) for information coding, the Hodgkin-Huxley model is used to obtain a more biological plausible description of action potential with plenty of ionic activities. To validate the effectiveness of the proposed model, we apply the model to the unmanned aerial vehicle (UAV) autonomous learning process in a 3-D environment. Experimental results show that our model is able to give the UAV the ability of free exploration in the environment and has comparable learning speed as the Q learning algorithm, while the major advances for our model is that it is with solid cognitive and neural basis.

Posted in: Applications of reinforcement learning to robots, Psycho-physiological bases of engineering , Tagged: