Juan López-Coronado

Neuro-Inspired Spike-Based Motion: From Dynamic Vision Sensor to Robot Motor Open-Loop Control through Spike-VITE

In this paper we present a complete spike-based architecture: from a Dynamic Vision Sensor (retina) to a stereo head robotic platform. The aim of this research is to reproduce intended movements performed by humans taking into account as many features as possible from the biological point of view. This paper fills the gap between current spike silicon sensors and robotic actuators by applying a spike processing strategy to the data flows in real time. The architecture is divided into layers: the retina, visual information processing, the trajectory generator layer which uses a neuroinspired algorithm (SVITE) that can be replicated into as many times as DoF the robot has; and finally the actuation layer to supply the spikes to the robot (using PFM). All the layers do their tasks in a spike-processing mode, and they communicate each other through the neuro-inspired AER protocol. The open-loop controller is implemented on FPGA using AER interfaces developed by RTC Lab. Experimental results reveal the viability of this spike-based controller. Two main advantages are: low hardware resources (2% of a Xilinx Spartan 6) and power requirements (3.4 W) to control a robot with a high number of DoF (up to 100 for a Xilinx Spartan 6). It also evidences the suitable use of AER as a communication protocol between processing and actuation.

Hyper RBF model for accurate reaching in redundant robotic systems

In this paper, a solution based on hyper radial basis functions networks (HRBF) for learning inverse kinematics in redundant robots is presented. This model has been implemented in two different visuo-motor robotic platforms for reaching and grasping applications. The obtained results allow to verify the robustness and accuracy capabilities of this neural model for reaching and tracking objects as well as to give a solution when redundant robots are considered. Therefore, the invariance of the proposed visuo-motor architecture for different arm-head relative configurations is demonstrated.

Safe-/spl mu/ARTMAP: a new solution for reducing category proliferation in fuzzy ARTMAP

/spl mu/ARTMAP is a neural network architecture that addresses the category proliferation problem present in fuzzy ARTMAP, by encouraging the creation of large hyperboxes. However, under certain characteristics of the classification task, this principle can be inadequate, namely if some classes have their patterns distributed in several isolated regions, far apart in the input space. Here we propose Safe-/spl mu/ARTMAP, a generalization of /spl mu/ARTMAP that limits the growth of a category in response to a single pattern, so that large hyperboxes are not created under these conditions. Experimental results confirm that the performance improves in some synthetic and real world tasks.

AER Neuro-Inspired interface to Anthropomorphic Robotic Hand

Address-event-representation (AER) is a communication protocol for transferring asynchronous events between VLSI chips, originally developed for neuro-inspired processing systems (for example, image processing). Such systems may consist of a complicated hierarchical structure with many chips that transmit data among them in real time, while performing some processing (for example, convolutions). The information transmitted is a sequence of spikes coded using high speed digital buses. These multi-layer and multi-chip AER systems perform actually not only image processing, but also audio processing, filtering, learning, locomotion, etc. This paper present an AER interface for controlling an anthropomorphic robotic hand with a neuro-inspired system.

A Neural Tactile Architecture Applied to Real-time Stiffness Estimation for a Large Scale of Robotic Grasping Systems

This paper presents a model for solving the problem of real-time neural estimation of stiffness characteristics for unknown objects. For that, an original neural architecture is proposed for a large scale robotic grasping systems applied for unknown object with unspecified stiffness characteristics. The force acquisition is based on tactile information from force sensors in robotic manipulator. The proposed model has been implemented on a robotic gripper with two parallel fingers and on a one d.o.f. robotic finger with opponent artificial muscles and angular displacements. This self-organized model is inspired of human biological system, and is carried out by means of Topographic Maps and Vector Associative Maps. Experimental results demonstrate the efficiency of this new approach.

A neural model for visual-tactile-motor integration in robotic reaching and grasping tasks

This paper presents a neural model to solve the visual-tactile-motor coordination problem in robotic applications. The proposed neural controller is based on the VAMC (Vector Associative Map) model. This algorithm is based on the human biological system and has the ability of learning the mapping that establishes the relationship between the spatial and the motor coordinates. These spatial inputs are composed of visual and force parameters. The LINCE stereohead carries out a visual detection process, detecting the positions of the object and of the manipulator. The artificial tactile skins placed over the two fingers of the gripper measure the force distribution when an object is touched. The neural controller has been implemented for robotic operations of reaching and object grasping. The reaching process is fed back in order to minimize the Difference Vector (DV) between the visual projections of the object and the manipulator. The stable grasping task processes the force distribution maps detected in the contact with the two surfaces of the gripper, in order to direct the object into the robotic fingers. Experimental results have demonstrated the robustness of the model and the accuracy of the final pick-and-place process.

Design of an anthropomorphic finger using shape memory alloy springs

We present the design of an anthropomorphic finger prototype. In this artificial finger, the actuators are electric pistons, whose main component is a shape memory alloy (SMA) spring. The artificial finger presents three independent degrees of freedom (DOF) for the metacarpophalangeal, proximal interphalangeal, and distal interphalangeal joints, respectively. The paper outlines the kinematic and structural characteristics of the finger. The main goal pursued during the development of the finger has been that of designing a small and lightweight dextrous gripper with anthropomorphic kinematics, which could be easily ported and installed even on small robot hands. We propose to use a physical anthropomorphic finger to demonstrate and validate a neural controller based on biological models. The neural controller applies a strategy of trajectory control using the vector integration to endpoint (VITE) model, which exhibits key kinematic properties of human movements, including asymmetric bell-shaped velocity profiles. The VITE model is used to compute the desired joint movement trajectories by smoothly interpolating between initial and final muscle length commands for the antagonist muscles involved in the movement. The rate of interpolation is controlled by the product of a difference vector which continuously computes the difference between the desired and present position of the finger, and a volitional movement gating signal. Experimental performance results in the time domain are presented, and directions for future research are discussed.

A neural controller for a robotic hand with artificial tactile skins in grasping tasks

In this paper, artificial tactile skins have been applied to a grasping task which require a certain precision in the determination of the object contact position with the surfaces and the pressure exercised in each point of the same. This design allows to process the force distribution maps in order to do precise maneuvers of grasping.

Position control based on static neural networks of anthropomorphic robotic fingers

A position neurocontroller for robot manipulators with a tendon-driven transmission system has been developed allowing to track desired trajectories and reject external disturbances. The main problem to control tendons proceeds from the different dimensions between the joint and the tendon spaces. In order to solve this problem we propose a static neural network in cascade with a torque resolutor. The position controller is built as a parametric neural network by using basis functions obtained directly from the finger structure. This controller insure that the tracking error converges to zero and the weights of the network are bounded. The implementation has been improved partitioning the neural network into subnets and using the Kronecker product. Both control and weight updating laws have been designed by means of a Lyapunov energy function. In order to improve the computational efficient of the neural network, this has been split up into subnets to compensate inertial, Coriolis/centrifugal and gravitational effects. The NN weights are initialised at zero and tuned on-line with no ”off-line learning phase”. This scheme has been applied to an anthropomorphic robotic finger with a transmission system based on tendons.

A modular neural network linking Hyper RBF and AVITE models for reaching moving objects

In this paper, the problem of precision reaching applications in robotic systems for scenarios with static and non-static objects has been considered and a solution based on a modular neural architecture has been proposed and implemented. The goal of this solution is to combine robustness and capability mapping trajectories from two biologically plausible neural network sub-modules: Hyper RBF and AVITE. The Hyper Basis Radial Function (HypRBF) neural network solves the inverse kinematic in redundant robotic systems, while the Adaptive Vector Integration to End-Point (AVITE) visuo-motor neural model quickly maps the difference vector between current and desired position in both spatial (visual information) and motor coordinates (propioceptive information). The anthropomorphic behaviour of the proposed architecture for reaching and tracking tasks in presence of spatial perturbations has been validated over a real arm-head robotic platform.