Joint Position Sensors Research Articles

Unknown Slope-Oriented Research on Model Predictive Control for Quadruped Robot

There are many undulating terrains in the wild environment. In order to realize the adaptive and stable walking of quadruped robots on unknown sloped terrain, a slope-adaptability model predictive control (SAMPC) algorithm is proposed in this work. In the absence of external vision sensors, the orientation and inclination of the slope are estimated based on the joint position sensors and inertial measurement units (IMU). In an effort to increase the stability margin, the adaptive algorithm adjusts the attitude angle and the touch-down point of the swing leg. To reduce the slipping risk, a nonlinear control law is designed to determine the friction factor of the friction cone constraint from the inclination of the slope. We validate the effectiveness of the framework through a series of simulations. The automatic smooth transition from the flat to the unknown slope is achieved, and the robot is capable of walking in all directions on the slope. Notably, with reference to the climbing modal of blue sheep, the robot successfully climbed a 42.4° slope, proving the ultimate ability of the proposed framework.

NARX Neural Network for Safe Human–Robot Collaboration Using Only Joint Position Sensor

Background: Safety is the very necessary issue that must be considered during human-robot collaboration in the same workspace or area. Methods: In this manuscript, a nonlinear autoregressive model with an exog-enous inputs neural network (NARXNN) is developed for the detection of collisions between a manipulator and human. The design of the NARXNN depends on the dynamics of the manipulator’s joints and considers only the signals of the position sensors that are intrinsic to the manipulator’s joints. Therefore, this network could be applied and used with any conventional robot. The data used for training the designed NARXNN are generated by two experiments considering the sinusoidal joint motion of the manipulator. The first experiment is executed using a free-of-contact motion, whereas in the second experiment, random collisions by human hands are performed with the robot. The training process of the NARXNN is carried out using the Levenberg–Marquardt algorithm in MATLAB. The evaluation and the effectiveness (%) of the developed method are investigated taking into account different data and conditions from the used data for training. The experiments are executed using the KUKA LWR IV manipulator. Results: The results prove that the trained method is efficient in estimating the external joint torque and in correctly detecting the collisions. Conclusions: Eventually, a comparison is presented between the proposed NARXNN and the other NN architectures presented in our previous work.

Neural networks design and training for safe human-robot cooperation

In the present paper, a neural network (NN) is proposed for detecting the human-manipulator collisions. Because of that purpose, three types of NNs are designed and trained named as; multilayer feedforward, cascaded forward, and recurrent NNs. The NN is designed based on the joints’ dynamics of the manipulator. In addition, the NN is trained using a dataset with and without the collisions using the algorithm of Levenberg-Marquardt to detect the collisions happened with the manipulator. During the design of the NN, only the intrinsic joint position sensor of the robot is used which enable the proposed method to be applied to any robot. The three designed neural networks are compared quantitatively and qualitatively to each other. Furthermore, the proposed method was evaluated with the KUKA LWR manipulator using one joint motion. The results of the designed systems achieve high effectiveness in detecting the collisions between human and robot.

Human–robot collisions detection for safe human–robot interaction using one multi-input–output neural network

In this paper, a multilayer feedforward neural network-based approach is proposed for human–robot collision detection taking safety standards into consideration. One multi-output neural network is designed and trained using data from the coupled dynamics of the manipulator with and without external contacts to detect unwanted collisions and to identify the collided link using only the intrinsic joint position and torque sensors of the manipulator. The proposed method is applied to the collaborative robots, which will be very popular in the near future, and is implemented and evaluated in 3D space motion taking into account the effect of the gravity. KUKA LWR manipulator is an example of the collaborative robots, and it is used for doing the experiments. The experimental results prove that the developed system is considerably efficient and very fast in detecting the collisions in the safe region and identifying the collided link along the entire workspace of the three-joint motion of the manipulator. Separate/uncoupled neural networks, one for each joint, are also designed and trained using the same data, and their performance is compared with the coupled one.

Neural Network Design for Manipulator Collision Detection Based Only on the Joint Position Sensors

SUMMARYIn this paper, a multilayer feedforward neural network (NN) is designed and trained, for human–robot collisions detection, using only the intrinsic joint position sensors of a manipulator. The topology of one NN is designed considering the coupled dynamics of the robot and trained, with and without external contacts, by Levenberg–Marquardt algorithm to detect unwanted collisions of the human operator with the manipulator and the link that is collided. The proposed approach could be applied to any industrial robot, where only the joint position signals are available. The designed NN is compared quantitatively and qualitatively with an NN, where both the intrinsic joint position and the torque sensors of the manipulator are used. The proposed method is evaluated experimentally with the KUKA LWR manipulator, which is considered as an example of the collaborative robots, using two of its joints in a planar horizontal motion. The results illustrate that the developed system is efficient and fast to detect the collisions and identify the collided link.

Retina-Based Pipe-Like Object Tracking Implemented Through Spiking Neural Network on a Snake Robot.

Vision based-target tracking ability is crucial to bio-inspired snake robots for exploring unknown environments. However, it is difficult for the traditional vision modules of snake robots to overcome the image blur resulting from periodic swings. A promising approach is to use a neuromorphic vision sensor (NVS), which mimics the biological retina to detect a target at a higher temporal frequency and in a wider dynamic range. In this study, an NVS and a spiking neural network (SNN) were performed on a snake robot for the first time to achieve pipe-like object tracking. An SNN based on Hough Transform was designed to detect a target with an asynchronous event stream fed by the NVS. Combining the state of snake motion analyzed by the joint position sensors, a tracking framework was proposed. The experimental results obtained from the simulator demonstrated the validity of our framework and the autonomous locomotion ability of our snake robot. Comparing the performances of the SNN model on CPUs and on GPUs, respectively, the SNN model showed the best performance on a GPU under a simplified and synchronous update rule while it possessed higher precision on a CPU in an asynchronous way.

Detecting slipping-like perturbations by using adaptive oscillators.

This study introduces a novel algorithm to detect unexpected slipping-like perturbations based on the comparison between actual leg joint angles and those predicted by a pool of adaptive oscillators. The approach grounds on the hypothesis that during postural transitions, the difference between these datasets diverges and can early signal that the dynamic balance is challenged. To test this hypothesis, leg joint angles of twelve healthy young participants were recorded while undergoing four different perturbations delivered during steady locomotion. Joint angles were estimated after spanning the whole domain of the adaptive oscillator dynamics. Results confirmed that the implemented strategy allows to early detect a postural transition induced by a slipping-like perturbation: the best performance is represented by a mean detection time ranging between 150 and 250ms and a low rate (lower than 10%) of false alarms. On the whole, the proposed approach is efficient even if it is based on a quite simple threshold-based algorithm. Moreover, it does not need any falling-based training before being implemented, is not computationally heavy, and is not subject dependent. Finally, since it is based on leg joint angles, it appears well suited to be implemented in lower-limb orthoses/prostheses already equipped with joint position sensors.

Adaptive Filtering for Robust Proprioceptive Robot Impact Detection Under Model Uncertainties

In the context of safe human–robot physical interaction, this paper introduces a new method for the detection of dynamic impacts of flexible-joint robot manipulators with their environment. The objective is to detect external impacts applied to the robot using only proprioceptive information with maximal sensitivity. Several model-based detection methods in robotics are based on the difference, called residual, between the estimated and the actual applied torques. Sensitivity of such methods can be limited by model uncertainties that originate either from errors on experimentally identified model parameters, possibly varying with the operating conditions, or the use of simplified models, which results in a residual dependence on the robot’s state. The main contribution of this paper consists of a new adaptive residual evaluation method that takes into account this dependence, which otherwise can lead to a tradeoff between sensitivity and false alarm rate. The proposed approach uses only proprioceptive motor-side measurements and does not require any additional joint position sensors or force/torque sensors. Dynamic effects of a collision on the residual are isolated using bandpass filtering and comparison with a state-dependent dynamic threshold. Adaptive online estimation of filter coefficients avoids the need for extensive experiments for parametric model identification. Experimental evaluation on the CEA backdrivable ASSIST robot arm illustrates the enhancement of the detection sensitivity.

Development of a Low-Cost Inspection Arm to Map the Available Workspace Within the Abdominal Cavity

Development of a Low-Cost Inspection Arm to Map the Available Workspace Within the Abdominal Cavity

Observer-Based Dynamic Control of an Underactuated Hand

The purpose of this paper was to construct a velocity observer based on the dynamic model and realize accurate dynamic curve and force control. Curve fitting with the observer obtained precise velocity signals. Compared with PID and factored moment methods, it decreased the fitting errors a lot and achieved ideal results. Compensated with the inverse dynamic equation, the force-based impedance control with the observer could not only realize accurate force tracking, but achieve finger dynamic control by the combination of curve fitting and force tracking. Furthermore, a static grasp model was established for appropriate force distribution. The finger could grasp slippery, fragile, comparatively heavy and large objects like an egg with only base joint torque and position sensors, which illustrated that the hand could accomplish difficult tasks by using the static grasp model and dynamic control.

Self-calibrated robotic manipulator force observer

In the robotic manipulation context, end-effector contact forces may be difficult to measure mainly due to the tool dynamic interferences such as the inertial forces. In this paper, a whole methodology is proposed to estimate these forces. The new approach is based on a sensor fusion technique that integrates the information of a wrist force sensor, of a 3D accelerometer placed at the robot tool and the joint position sensors measurements. The proposed methodology not only offers a suitable estimator in terms of response and filtering, but also presents a self-calibrating feature that allows an easy integration into any industrial setup. To experimentally validate the performance of the proposed methodology, two different industrial manipulators were used: an ABB robot and a Stäubli robot, both with open control system architectures. An impedance control scheme was used as force/position control law to demonstrate the need and results of the proposed calibration result.

Neural virtual sensors for terrain adaptation of walking machines

When walking in realistic conditions, accurate, reliable sensorial information is critical to ensure the safe operation of legged robots. That means a large number of sensors, cabling, and electronic systems must be used, complicating the robot. On the other hand, the great complexity of the hardware of walking machines is one of the main obstacles preventing the introduction of this kind of vehicle in real applications; consequently this hardware should be simplified. These antagonistic requirements can be reconciled by the use of what are called virtual sensors. This paper addresses the design of virtual sensors for terrain adaptation developed with the aims of simplifying the hardware of the walking machine or increasing the reliability of the sensorial information available. These virtual sensors are based on neural networks and can estimate the forces exerted by the feet from data extracted from joint-position sensors, which are mandatory in all robotic systems. The force estimates are used to detect foot/ground contact. Some experiments carried out with the SILO4 walking robot are reported to prove the efficacy of this method. © 2005 Wiley Periodicals, Inc.

Passivity-Based Interaction Controller and Observer for Robot Manipulators

This work presents the design of a passivity-based controller and observer for a robot manipulator interacting with a compliant environment. Regulation of the contact force along the constrained direction and tracking of the end-effector position along the unconstrained directions is achieved; semiglobal exponential stability is proved. The control scheme is implemented in the joint space, and experimental results are illustrated for an open control architecture industrial robot which is endowed with joint position sensors and wrist force sensor, but no joint velocity sensors.

Real Time Implementation of the Impedance Control of a Robot Arm

The implementation of the impedance controller requires exact linearization in operational space which is achieved by computed torque approach. In the paper a perturbation observer has used in one case to obtain the complete linearization torque and in another case to obtain only the compensating torque for unmodelled dynamic terms. This perturbation observer requires the joint acceleration feedback signal. The calculation by numerical derivation of joint position assumes that high resolution joint resolvers are available. Joint acceleration can be measured using accelerometers. Combining the end-effector accelerometers and joint position sensors can give acceleration estimate with higher bandwidth than each separately. Extended Kalman Filter was selected as a tool signal fusion. Operational space control was verified experimentally using fusion scheme based on an Extended Kalman Filter.

Attitude and position control for non-rigid walking machines

This paper deals with mechanical problems that appear in realistic walking machines. Flexion in links and joints produces a change in body attitude and position which must be corrected. This phenomenon is difficult to model, therefore, sensors must be used. This paper proposes an attitude/position control strategy for a walker that uses a class of periodic gaits known as wave-crab gaits. To perform this controlling action, sensors such as contact sensors, inclinometers and joint position sensors are employed. The main point of this paper is to analyze the phenomenon and to describe the proposed algorithms and the experimental results obtained with a quadruped walking machine named RIMHO.

Dynamic contact sensing by flexible beam

This paper discusses a new dynamic sensing system capable of detecting the contact point between a flexible beam and an object. The proposed sensing system, named dynamic antenna, is simply composed of an insensitive flexible beam, a torque sensor, a joint position sensor, an actuator, and a payload at the tip end of the beam. The contact point can be detected through estimation of the oscillation frequencies of the beam in contact with the object. First, a dynamic model of the sensor is derived. Next, it is shown that information of the fundamental and the second natural frequencies is sufficient for unique determining of the contact point if the beam has uniform mass and stiffness distribution. In practical realization, the fundamental and the second natural frequencies of the beam in contact with the object are extracted from the torque sensor measurements with the use of the maximum entropy method. Then, the frequencies are mapped into the contact-point coordinate. Extraction of the frequencies and mapping them into the contact point constitute a sensing strategy which is tested under experiment.

Attitude and position control method for realistic legged vehicles

Flexion in links and joints produces a change in both attitude and position of the body which must be corrected. Because this phenomenon is difficult to model, this paper proposes an attitude/position control strategy based on sensors such as contact sensors, inclinometers, and joint position sensors. This method is aimed at a class of periodic gaits known as wave gaits. The main point of this paper is to analyze this phenomenon and to describe the proposed algorithms and the experimental results obtained with a quadruped walking machine named RIMHO.

Control of cooperative multiple flexible joint robots

The problem of controlling multiple flexible joint robots during cooperative manipulation of a rigidly grasped common load (denoted as the CMFJR problem) is addressed. The dynamic model of the CMFJR shows the interaction between the manipulators in the closed chain is analogous to a single manipulator interacting with frictionless (algebraic constraint) surface, while in motion along the surface. It is assumed that each flexible joint robot is equipped with joint velocity and position sensors and also actuator mounted position and velocity sensors. It is shown that it is not necessary to employ force sensors mounted on the robot wrist in order to track a desired load trajectory and an internal force trajectory. The transient response of the position error and the internal force error can be arbitrarily assigned even though the force sensors are not employed. >

Tendon organ afferents in the knee joint nerve of the cat

This article is concerned with the question of whether some of the slowly adapting responses recorded in the mid-range of joint position from the posterior articular nerve (PAN) of the cat hindlimb and which are therefore candidates for joint position sensors, are in fact coming from muscle receptors of the popliteus muscle. Evidence is provided that in addition to afferents of primary endings of spindles, secondary endings of spindles and tendon organs are represented in PAN as well. Removal of the fleshy part of the popliteus muscle does not reliably remove muscle afferent contamination in PAN.

Computer Control of a 6 Degrees-of-Freedom Articulated Manipulator

Construction and control problems of a novel, 6 axis, serial robot arm are discussed. Newly developped, miniaturized drives are used in the mechanics of the arm. Joint position and force-torque sensors give information about the state of the arm and its environment. A kinematic control model is set up and algorithms are found to transform hand trajectories into joint coordinates and to control the joints in a coordinated manner. A new method is used for defining the workspace and motions of the manipulator: parts and robot actions are described in a graphic modelling system.