Duc Trong Tran

Central pattern generator based reflexive control of quadruped walking robots using a recurrent neural network

This paper presents a novel Central Pattern Generator (CPG) model for controlling quadruped walking robots. The improvement of this model focuses on generating any desired waveforms along with accurate online modulation. In detail, a well-analyzed Recurrent Neural Network is used as the oscillators to generate simple harmonic periodic signals that exhibit limit cycle effects. Then, an approximate Fourier series is employed to transform those mentioned simple signals into arbitrary desired outputs under the phase constraints of several primary quadruped gaits. With comprehensive closed-form equations, the model also allows the user to modulate the waveform, the frequency and the phase constraint of the outputs online by directly setting the inner parameters without the need for any manual tuning. In addition, an associated controller is designed using leg coordination Cartesian position as the control state space based on which stiffness control is performed at sub-controller level. In addition, several reflex modules are embedded to transform the feedback of all sensors into the CPG space. This helps the CPG recognize external disturbances and utilize inner limit cycle effect to stabilize the robot motion. Finally, experiments with a real quadruped robot named AiDIN III performing several dynamic trotting tasks on several unknown natural terrains are presented to validate the effectiveness of the proposed CPG model and controller.

Sensing and gait planning of quadruped walking and climbing robot for traversing in complex environment

In this paper, a general study on improving adaptability of quadruped walking and climbing robot in complex environment is presented. First, a sensing system composed of range and gyroscope sensors in a novel arrangement is developed. By combining the sensing signals and the internal state of the robot, the surface geometry of the environment is sufficiently reconstructed in real-time. Secondly, a planning algorithm for the robot to overcome the reconstructed environment is conducted. Based on the reshaped surface, the planning algorithm not only provides the exact body trajectory and foot positions but also the adaptability of the robot in a specific environment. A method to improve the adaptability of the walking and climbing robot is also introduced. Thanks to the adherent ability of the robot, the center of gravity of the robot is allowed to move outside the support polygon to increase the reach-ability of the next swing leg. Finally, the effectiveness of the proposed approach is verified by the performances of the experiments in complex environments using a quadruped walking and climbing robot named MRWALLSPECT IV.

Development of wall climbing robotic system for inspection purpose

In this paper, we introduce a wall climbing robotic system for visual inspection of man-made structures. The adhesion mechanism of our system consists of an impeller and two-layer suction seals which provide sufficient adhesion forces for supporting the robot body on the non-smooth vertical wall and horizontal ceiling by generating pressure difference between the inside of the pressure chamber and the ambient environment. A comprehensive study is performed on the dynamic fluid modeling of the adhesion mechanism and the adhesion force is controlled by adjusting the pressure inside of the chamber. In addition, stable differential-driving locomotion on non-smooth surface are achieved by adapting a suspension mechanism for each wheel. A wall climbing robot, called LARVA, is successfully developed and its effectiveness of locomotion is verified with experiments on non-smooth vertical wall and horizontal ceiling surface.

Improving traversability of quadruped walking robots using body movement in 3D rough terrains

This paper presents a study on improving the traversability of a quadruped walking robot in 3D rough terrains. The key idea is to exploit body movement of the robot. The position and orientation of the robot are systematically adjusted and the possibility of finding a valid foothold for the next swing is maximized, which makes the robot have more chances to overcome the rough terrains. In addition, a foothold search algorithm that provides the valid foothold while maintaining a high traversability of the robot, is investigated and a gait selection algorithm is developed to help the robot avoid deadlock situations. To explain the algorithms, new concepts such as reachable area, stable area, potential search direction, and complementary kinematic margin are introduced, and the effectiveness of the algorithms is validated via simulations and experiments.

A new method in modeling Central Pattern Generators to control quadruped walking robots

Aiming to real easy application in several quadruped robot platforms, this paper introduces a new method of modeling central pattern generators (CPG) to control quadruped locomotion. Not only can this new model generate all the primary gaits of quadrupeds stably with limit cycle effect, but it also has the ability of tuning the periodic outputs with arbitrary waveforms. The core idea is to combine strong points of two mathematical tools: Fourier series and Recurrent neural networks. In addition, a new biomimetic controller is also introduced using the proposed CPG model and several reflex modules. Finally, dynamic simulations are performed to validate the efficiency of the proposed controller.

Body Workspace of Quadruped Walking Robot and its Applicability in Legged Locomotion.

This paper discusses on determination of the workspace of the body of a quadruped walking robot, called “body workspace”, and its applicability in legged locomotion. The body workspace represents the set of all valid body configurations for a next step by considering three constraints of a body position: existence of the inverse kinematic solutions, reach-ability of the next swing leg to the next desired foothold, and static equilibrium of the robot when the next swing leg is lifted. The space contains all the body positions that ensure the existence of inverse kinematic solutions, is calculated in the first. Then, a subspace inside the determined space that allows the robot to reach the next desired foothold is analyzed. Finally, the workspace is obtained by excluding all the positions inside the subspace that do not ensure the equilibrium of the robot when the next swing leg is lifted. Therefore, the workspace shows all possible solutions for choosing the next body configuration of a given static walking problem. It is significant in improving the robot’s performances since moving body takes an intrinsic role in static walking, besides swinging a leg. The algorithm runs fast in real-time because it is a pure geometric method. The body workspace of a quadruped walking robot is visualized to help the understanding of the algorithm. In addition, applications of using the body workspace in improving the robot’s ability are presented to show potential applicability of the workspace.

A gait transition algorithm based on hybrid walking gait for a quadruped walking robot

This paper presents a quasi-dynamic gait, called Hybrid Walking Gait, and a new gait transition algorithm for a quadruped walking robot. The Hybrid Walking gait reduces the steps of a generic walking gait with primitive foot trajectory generation using some of parameters easily defined. It shows great improvements over existing ones in terms of higher mobility, less complexity to define the motion, and smooth body movements that affect to the stability of the robot. The Gait Transition pattern generated with the Hybrid Walking Gait guarantees stability as good as that of a traditional walking gait and high mobility such as the dynamic trot gait. We perform experiments with a quadruped robot called “Artificial Digitigrade for Natural Environments Version III”, and validate the effectiveness of our proposed gait patterns over several types of terrains.

Control of a quadruped robot with enhanced adaptability over unstructured terrain

Improvement of the adaptability of a quadruped robot in rough terrain is studied in this paper. First, the position and posture of the body of the robot are adjusted to maximize the number of choices for foot placement of the next swing leg. The more choices the robot has to select the next suitable foothold, the better it will be to cope with rough terrain. Second, an effective foothold search algorithm is developed. The foothold search algorithm not only tries to find a valid foothold but also tries to maintain high adaptability of the robot. For implementing the algorithm, some new concepts such as potential swing direction, complementary kinematic margin and elliptical set of candidate footholds are also proposed. The effectiveness of this procedure in improving the adaptability of a quadruped robot moving in challenging terrain is verified in both simulation and experiment.

An on-line gravity estimation method using inverse gravity regressor for robot manipulator control

When a robotic manipulator is controlled, computing gravity force of the robot is the primary issue. Exact model parameters are not easy to be known in the practical robot system due to the uncertainty of the robot dynamics. Hence, the gravity force is presented by a combination of gravity regressor and robot dynamic parameters and is compensated by the estimation of uncertain robot dynamic parameters. Previous researches conducted estimation by using transpose of gravity regressor and full form of dynamic parameters which is not general form however this paper estimates the gravity force using the generalized gravity regressor which is regardless of the dimension and structure of the robot under the quasi-static state. Once the estimation is completed, the estimated value can be used to compute the gravitational force and control the robot. It is shown that the generalized decomposition of gravity regressor and estimation process. The results are validated through an experiment by implementing the algorithm on an upper-body dual arm robot.

Quadruped locomotion: Dynamic gait control & design optimization

In this presentation, we introduce two majored contents including novel algorithms in control of dynamic locomotion of quadruped robots on unknown rough terrains and the method towards the optimization solutions in running legged robots. The control strategy was designed based on two different approaches. The first one uses the biologically inspired artificial Central pattern generators as the controller core whereas the second approach proposed a new method to approximate the leg motion using 3D spring-damper system model. Both two approaches were thoroughly studied in simulations and real experiments in a quadruped robot named AiDIN III. Besides, we present a practical procedure to solve the optimization problem of running robot design. The core idea is to use numerical tools to indirectly search for the parameters which results in not only self-stable motion but also optimized hardware structure.