Search

The correct and robust recognition of traffic signs is indispensable to self-driving vehicles and driver-assistant systems. In this work, we propose and evaluate two network architectures for multi-expert decision systems that we test on a challenging Traffic Sign Recognition Benchmark dataset. The decision systems implement individual experts in the form of deep convolutional neural networks (CNNs). A gating network CNN acts as final decision unit and learns which individual expert CNNs are likely to contribute to an overall meaningful classification of a traffic sign. The gating network then selects the outputs of those individual expert CNNs to be fused to form the final decision. In this work we study the advantages and challenges of the proposed multi-expert architectures that in comparison to other network architectures allow for parallel training of individual experts with reduced datasets. Under the challenging conditions introduced by the benchmark dataset, the demonstrated multi-expert decision systems achieve a recognition performance that is superior to those of humans: with an accuracy of 99.10%, when training experts with the complete dataset and 98.94%, when individual experts are only trained with 36% of the training samples. Overall, our approach ranked fourth on the list of the applied approaches proposed for the German traffic sign Recognition Benchmark (GTSRB) dataset.

This paper develops a new method for coevolution, named Fitness-Diversity Driven Coevolution (FDDC). This approach builds on existing methods by a combination of a (predator-prey) Coevolutionary Genetic Algorithm (CGA) and novelty search. The innovation lies in replacing the absolute novelty measure with a relative one, called Fitness-Diversity. FDDC overcomes problems common in both CGAs (premature convergence and unbalanced coevolution) and in novelty search (construction of an archive). As a proof of principle, Spring Loaded Inverted Pendulums (SLIPs) are coevolved with 2D-terrains the SLIPs must learn to traverse.

Bioinspired legged locomotion comprises different aspects, such as (i) benefiting from reduced complexity control approaches as observed in humans/animals, (ii) combining embodiment with the controllers and (iii) reflecting neural control mechanisms. One of the most important lessons learned from nature is the significant role of compliance in simplifying control, enhancing energy efficiency and robustness against perturbations for legged locomotion. In this research, we investigate how body morphology in combination with actuator design may facilitate motor control of leg function. Inspired by the human leg muscular system, we show that biarticular muscles have a key role in balancing the upper body, joint coordination and swing leg control. Appropriate adjustment of biarticular spring rest length and stiffness can simplify the control and also reduce energy consumption. In order to test these findings, the BioBiped3 robot was developed as a new version of BioBiped series of biologically inspired, compliant musculoskeletal robots. In this robot, three-segmented legs actuated by mono- and biarticular series elastic actuators mimic the nine major human leg muscle groups. With the new biarticular actuators in BioBiped3, novel simplified control concepts for postural balance and for joint coordination in rebounding movements (drop jumps) were demonstrated and approved.

Imagine a world in which our furniture moves around like legged robots, interacts with us, and changes shape and function during the day according to our needs. This is the long term vision we have in the Roombots project. To work towards this dream, we are developing modular robotic modules that have rotational degrees of freedom for locomotion as well as active connection mechanisms for runtime reconfiguration. A piece of furniture, e. g. a stool, will thus be composed of several modules that activate their rotational joints together to implement locomotor gaits, and will be able to change shape, e. g. transforming into a chair, by sequences of attachments and detachments of modules.In this article, we firstly present the project and the hardware we are currently developing. We explore how reconfiguration from a configuration A to a configuration B can be controlled in a distributed fashion. This is done using meta-modules-two Roombots modules connected serially-that use broadcast signals and connections to a structured ground to collectively build desired structures without the need of a centralized planner.We then present how locomotion controllers can be implemented in a distributed system of coupled oscillators-one per degree of freedom-similarly to the concept of central pattern generators (CPGs) found in the spinal cord of vertebrate animals. The CPGs are based on coupled phase oscillators to ensure synchronized behavior and have different output filters to allow switching between oscillations and rotations. A stochastic optimization algorithm is used to explore optimal CPG configurations for different simulated Roombots structures.

This study aims at providing a control-learning framework capable of generating optimal locomotion patterns for the modular robots. The key ideas are firstly to provide a generic control structure that can be well-adapted for the different morphologies and secondly to exploit and coevolve both morphology and control aspects. A generic framework combining robot morphology, control and environment and on the top of them optimization and evolutionary algorithms are presented. The details of the components and some of the preliminary results are discussed. (C) Selection and peer-review under responsibility of FET11 conference organizers and published by Elsevier B.V.

Roombots (RB) are self-reconfigurable modular robots designed to study robotic reconfiguration on a structured grid and adaptive locomotion off grid. One of the main goals of this platform is to create adaptive furniture inside living spaces such as homes or offices. To ease the control of RB modules in these environments, we propose a novel and more natural way of interaction with the RB modules on a RB grid, called the Natural Roombots User Interface. In our method, the user commands the RB modules using pointing gestures. The user's body is tracked using multiple Kinects. The user is also given real-time visual feedback of their physical actions and the state of the system via LED illumination electronics installed on both RB modules and the grid. We demonstrate how our interface can be used to efficiently control RB modules on simple point-to-point grid locomotion and conclude by discussing future extensions.

The design of efficient locomotion controllers for arbitrary structures of reconfigurable modular robots is challenging because the morphology of the structure can change dynamically during the completion of a task. In this paper, we propose a new method to automatically generate reduced Central Pattern Generator (CPG) networks for locomotion control based on the detection of bio-inspired sub-structures, like body and limbs, and articulation joints inside the robotic structure. We demonstrate how that information, coupled with the potential symmetries in the structure, can be used to speed up the optimization of the gaits and investigate its impact on the solution quality (i.e. the velocity of the robotic structure and the potential internal collisions between robotic modules). We tested our approach on three simulated structures and observed that the reduced network topologies in the first iterations of the optimization process performed significantly better than the fully open ones.

Brain-Computer Interfaces (BCIs) are playing an increasingly important role in a broad spectrum of applications in health, industry, education, and entertainment. We present a novel, mobile and non-invasive BCI for advanced robot control that is based on a brain imaging method known as functional near-infrared spectroscopy (fNIRS). This BCI is based on the concept of \automated autonomous intention execution" (AutInEx), that is, the automated execution of possibly very complex actions and action sequences intended by a human through an autonomous robot.

Stable locomotion that tolerates parameter variations is an important feature for legged robots. In this paper we introduce a locomotion control framework for legged robots that combines the well-known spring-loaded inverted pendulum (SLIP) with active hip actuation through Central Pattern Generators. Using this framework we present studies suggesting that compliant in contrast to rigid hip actuation and the addition of a simple feedback scheme can highly enhance the robustness of the robot locomotion against parameter changes.