Legged Robots Research Papers - Academia.edu (original) (raw)

Boston Dynamics' robotic quadrupeds have achieved infamy and virality through a series of social media videos since 2008. In 2019 Boston Dynamics began commercial sale of 'Spot', a moving, sensing, networked robot dog. Spot has been designed to be a platform, which can be augmented with hardware payloads (e.g. sensors, robotic arm) and software to command Spot to conduct specific missions. In this paper we first trace the development of Spot and highlight the interest of the United States military in its development. This is followed by our text analysis of social media reactions to Boston Dynamics' quadrupeds, revealing public fascination as well as ongoing suspicion and dark humour about 'killer robots'. We then discuss how humanitarian applications, including in response to the COVID-19 pandemic, have been used as an opportunity to promote Spot and overcome public negativity. This is an example of a more general strategy advocates use to garner acceptance for autonomous robots in both civilian and military roles using humanitarian justifications: the robots 'save lives.' We conclude by discussing how Spot and other robot quadrupeds demonstrate the intertwining of humanitarian and military applications in the development, normalization and deployment of autonomous robots.

- by
- •
- Humanitarianism, Sentiment Analysis, Legged Robots, Military Robotics
- by Matt Haberland
- •
- Legged Robots
- by Tolga Karakurt
- •
- Mobile Robotics, Wireless Communications, Exoskeletons, Embedded Systems

This thesis presents advances in the state-of-the-art in legged locomotion through the de- velopment of bounding and galloping gaits as well as new modes of hybrid wheeled-leg modes of locomotion. Two four-legged running robots, Scout II and PAW, are examined, the latter of which is distinguished by actuated wheels at the ends of its legs.
First, hybrid modes of locomotion are demonstrated which use legs to dynamically reposition wheels at specific locations with respect to the body. These modes improve the stability and tire-wear of turning and braking manoeuvres and allow pitch-controlled slope ascent and descent in a wheeled-leg vehicle such as the PAW robot.
Second, through hip actuation, passive leg compliance and controlled wheel action it is possible to make the same vehicle run using a dynamically stable legged gait called the bound. Experimental evidence of this is presented and compared to similar experiments on the same robot with mechanically blocked wheels, a 3D simulation of the same, as well as bounding on a completely different quadrupedal robot, Scout II. While a casual observer finds no difference in blocked-wheel and active wheel control modes, detailed examination of the gaits reveals lower speeds and efficiency as well as decreased repeatability when the wheels are actively controlled.
A new method of forward speed control is presented for the bounding gait using liftoff, as opposed to touchdown, leg angles. The liftoff angle method of speed control is shown to be particularly suited to fine-tuning of certain gait performance indices.
Third, the underactuated bounding gait is extended to demonstrate, for the first time, that robotic galloping is possible and that it can be achieved in two underactuated quad- rupedal robots and with varying levels of decoupled control. In the Scout II robot the front leg pair and rear leg pairs function independently, while in the PAW robot galloping is achieved with no controlled coupling between any of the four legs. The rotary gallop gait demonstrated by both robots is characterized by a significant yaw component and is compared to another bound-derived turning gait which uses liftoff angles to produce yaw. In particular, the correspondence of lead leg to yaw direction in both cases is found to match results from biology. In contrast, while it is thought that animals pivot about their lead leg to turn, the rotary gallop demonstrated by these robots shows that yaw occurs primarily in the leg behind the lead leg.

- by James A Smith
- •
- Mobile Robotics, Biomechanics, Biology, Bio-Inspired Systems

Wheeled systems are energy efficient on prepared surfaces like roads and tracks. Legged systems are capable of traversing different terrains but can be lossy. At low speeds and on off-road surfaces, legged systems using dynamic walking can be efficient. Towards this objective, the dynamics of the walker needs to be modelled and controlled. This thesis presents analysis and experiments on the dynamics and control of a rimless wheel based mobile robot (Chatur ) in a category between wheeled and legged systems. It is effectively a 2D dynamic walker that serves as a platform for investigating inverted pendulum walking. A pulsed actuation torque is proposed for the system resulting in four torque regimes defined by the ratio of energy losses to available actuator torque. Five physical constraints that limit the choice of operating points of a generic inverted pendulum walker are expounded and location of optimal operating points is discussed. Chatur’s hardware design is elaborated and a control topology is proposed for pulsed actuation of the dual brushless dc (BLDC) motor driven platform with wheel synchronization.

- by Giuseppe Carbone and +1
- •
- Robotics, Legged Robots, Legged Locomotion

This article presents a summary of applications of chaos and fractals in robotics. Firstly, basic concepts of determin‐ istic chaos and fractals are discussed. Then, fundamental tools of chaos theory used for identifying and quantifying chaotic dynamics will be shared. Principal applications of chaos and fractal structures in robotics research, such as chaotic mobile robots, chaotic behaviour exhibited by mobile robots interacting with the environment, chaotic optimization algorithms, chaotic dynamics in bipedal locomotion and fractal mechanisms in modular robots will be presented. A brief survey is reported and an analysis of the reviewed publications is also presented.

- by Sajid Iqbal
- •
- Fractal Geometry, Mobile Robotics, Mathematical Biology, Swarm Intelligence

This paper studies the attitude dynamics and the control of quadruped robots using tail-like appendages during the flight phases of high speed locomotion. Inspiration and data are first obtained from cheetah’s fast galloping techniques. A two-body template is then used to simply describe the dynamics of a large body whose attitude is controlled by a rotating appendage. The equations of motion for a tail and a reaction wheel are given, while by employing cyclic coordinates, all possible reductions are performed to finally lead to the design of model-based controllers. A main contribution lies on the thorough discussion on the holonomy of the system, which only depends on the system’s geometry and the initial angular momentum. A comparison between a reaction wheel and a tail is also carried out, while basic steps and formulas are proposed for selecting the key parameters concerning the design of such systems. Finally, simulation results are presented in order to validate the methods proposed herein.

- by Konstantinos Machairas and +1
- •
- Legged Robots, Legged Locomotion
- by Juan Rodriguez
- •
- Robotics, Motion Planning, Legged Robots, Climbing

Wachter, S., Mittelstadt, B., & Floridi, L. (2017). Transparent, explainable, and accountable AI for
robotics. Science Robotics, 2(6), eaan6080.

- by Sandra Wachter and +1
- •
- Robotics, Computer Science, Algorithms, Artificial Intelligence

A series of experiments and simulations are presented here that provide insight into how an underactuated quadrupedal robot can gallop in a controlled manner simply by controlling touchdown and liftoff angles of the legs. Initial work resulted in a rotary gallop with significant yaw. On that basis, two hypotheses are presented and validated here to yield straight-line galloping. The first hypothesis is that the yaw can be controlled by adjusting the liftoff angle of a rear leg – specifically, the rear leg found on the inside radius of the turn. However, it is not guaranteed that the resulting gait will remain a gallop. Based on the validation of this first hypothesis and, having observed that animals can gallop in straight lines, a second hypothesis is put forward that, upon finding an initial stable gallop gait with non-negligible yaw, if sequential adjustments are made to both the touchdown and liftoff angles then it will be possible to zero the yaw while also providing for sufficient phase difference between successive toe contacts to classify the gait as a gallop. The hypotheses are confirmed with experiments showing a repeatable straightened gallop at 0.7 m/s.

- by James A Smith
- •
- Robotics, Mobile Robotics, Biomechanics, Running

— This article presents a complete formulation of the challenging task of stable humanoid robot omnidirectional walk based on the Cart and Table model for approximating the robot dynamics. For the control task, we propose two novel approaches: preview control augmented with the inverse system for negotiating strong disturbances and uneven terrain and linear model-predictive control approximated by an orthonormal basis for computational efficiency coupled with constraints for stability. For the generation of smooth feet trajectories, we present a new approach based on rigid body interpolation, enhanced by adaptive step correction. Finally, we present a sensor fusion approach for sensor-based state estimation and an effective solution to sensors' noise, delay, and bias issues, as well as to errors induced by the simplified dynamics and actuation imperfections. Our formulation is applied on a real NAO humanoid robot, where it achieves real-time onboard execution and yields smooth and stable gaits.

In this paper, we present a complete description of the hardware design and control architecture of our custom built quadruped robot, called the Stoch. Our goal is to realize a robust, modular, and a reliable quadrupedal platform, using which various locomotion behaviors are explored. This platform enables us to explore different research problems in legged locomotion, which use both traditional and learning based techniques. We discuss the merits and limitations of the platform in terms of exploitation of available behaviours, fast rapid prototyping, reproduction and repair. Towards the end, we will demonstrate trotting, bounding behaviors, and preliminary results in turning. In addition, we will also show various gait transitions i.e., trot-to-turn and trot-to-bound behaviors.

- by Dhaivat Dholakiya
- •
- Robotics, Mobile Robotics, Legged Robots, Quadruped Robots
- by Andres Valenzuela
- •
- Humanoid Robotics, Motion Planning, Legged Robots, Legged Locomotion

This paper describes a quadrupedal architecture assembly process using the modular robotic system Mecabot. Several possible topologies are considered, justifying the final design that allows using an active column. Based on this, a mathematical model of control is proposed to perform movements of displacement, open turn and rotation. The locomotion profiles for the first two movement modalities are bioinspired. For the rotation modality, a characteristic quadrupedal robot transition is used to allow the correct rotation without using a large number of degrees of freedom. The proposed control model was deployed in a robot tested on structured and unstructured terrains by measuring its speed as a function of the movement frequency variation. For the open turn modality, the turn radius was measured as a function of the offset variation. Based on the test results, the second Mecabot configuration with legs was finally obtained, complementing our research work on apodal (snake, wheel caterpillar) and hexapod configurations.

- by Visión Electrónica and +1
- •
- Legged Robots, Mobile Robots, Modular robots, Movil robots

Legged locomotion is a rapidly advancing area in robotics, yet still a large number of open questions exist. This work focuses on the foot-terrain interaction and its effect on the motion of a one-legged system. This interaction is usually tackled by disregarding some of the effects of ground deformation like permanent deformation and compaction. Inspired by other areas of engineering, an impact dynamics model is developed, allowing a more thorough study of the behavior during fast dynamic walking. This approach can be regarded as a viscoplastic one. The monopod controller presented in previous work is extended to cope with deformable terrains, based on energy dissipation considerations, without requiring the knowledge of the ground parameters. Simulation results prove the validity of the theory presented.

- by Iosif S Paraskevas and +2
- •
- Robotics, Control Systems Engineering, Dynamical Systems, Impact Mechanics

It is an important ability for any mobile robot to be able to estimate its posture and to gauge the distance it traveled. In this paper, we have addressed this problem in a dynamic quadruped robot by combining traditional state estimation methods with machine learning. We have designed and implemented a navigation algorithm for full body state (position, velocity, and attitude) estimation that uses no external reference but relies on multimodal proprioceptive sensory information only. The extended Kalman filter (EKF) was used to provide error estimation and data fusion from two independent sources of information: 1) strapdown mechanization algorithm processing raw inertial data and 2) legged odometry. We have devised a novel legged odometer that combines information from a multimodal combination of sensors (joint and pressure). We have shown our method to work for a dynamic turning gait, and we have also successfully demonstrated how it generalizes to different velocities and terrains. Furthermore, our solution proved to be immune to substantial slippage of the robot's feet.

- by Matej Hoffmann
- •
- Inertial navigation, Legged Robots, Dead Reckoning

Development and prototyping of robotic systems requires the involvement of many people and many hours of design, development, and cooperation; significant time and effort overhead is required for evaluating conceptual ideas in design, control and technology, and for bringing them fast into reality for testing. Based on the important advances of the last decade in hardware and software, a simple and low-cost framework and its underlying ideas are presented, with steps that aim at accelerating robotics research work in academia and industry. The framework’s functionality is validated and illustrated by two application examples concerning the control systems of a single-legged hopping robot and an instrumented treadmill. The software required to conduct the same experiments is provided, with the intention to help the reader reuse it in similar applications.

- by Konstantinos Machairas
- •
- Robotics, Legged Robots, Cyberphysical systems

In this research, an artificial chattering free adaptive fuzzy sliding mode control design and application to
uncertain robotic manipulator has proposed in order to design high performance nonlinear controller in the
presence of uncertainties. Regarding to the positive points in sliding mode controller, fuzzy logic controller
and adaptive method, the output has improved. Each method by adding to the previous controller has
covered negative points. The main target in this research is design of model free estimator on-line sliding
mode fuzzy algorithm for robot manipulator to reach an acceptable performance. Robot manipulators are
highly nonlinear, and a number of parameters are uncertain, therefore design model free controller using
both analytical and empirical paradigms are the main goal. Although classical sliding mode methodology
has acceptable performance with known dynamic parameters such as stability and robustness but there
are two important disadvantages as below: chattering phenomenon and mathematical nonlinear dynamic
equivalent controller part. To solve the chattering fuzzy logic inference applied instead of dead zone
function. To solve the equivalent problems in classical sliding mode controller this paper focuses on
applied fuzzy logic method in classical controller. This algorithm works very well in certain environment but
in uncertain or various dynamic parameters, it has slight chattering phenomenon. The system performance
in sliding mode controller and fuzzy sliding mode controller are sensitive to the sliding function. Therefore,
compute the optimum value of sliding function for a system is the next challenge. This problem has solved
by adjusting sliding function of the adaptive method continuously in real-time. In this way, the overall
system performance has improved with respect to the classical sliding mode controller. This controller
solved chattering phenomenon as well as mathematical nonlinear equivalent part by applied fuzzy
supervisory method in sliding mode fuzzy controller and tuning the sliding function.

- by FARZIN PILTAN and +3
- •
- Control Systems Engineering, Artificial Intelligence, Mechatronics, Fuzzy Logic

The revealed secrets of nature always led humans to their aspiring achievements. The fastest animal on land is Cheetah and similar robot has developed by engineers so far to attain a record speed of 20mph among legged robots. But in nature there are some insects those are far ahead of cheetah in speed with a unit of body length per second. Insects are small in their body size with legs usually countable from 4 to 12 or more. With more legs they can have more stability and can adapt to different terrain faster while walking. Six legged robot (hexapod) is generally expect to attain higher speed in terms of body length per second, since the nature has proof for it. Bio-inspired Central Pattern Generator (CPG) is in use for so far in robotic world to mimic the locomotion patterns of insects and other animals. Currently the hybrid controller of CPG and reflex is going on and this paper suggests a new architecture for the system. Neural Network modeled CPG acts as the motor neuron for each joint of the leg. In each instant a neural network models the gait of the robot by learning procedure from the reflex system. This is like the Central Nervous System (CNS) selecting gait of an animal according to the terrain that travels. CNS takes sensory feedback from eyes, force on each leg and body balance from cochlea to adapt the gait for current terrain. This paper in first place tries to simulate the gait patterns for a hexapod.

—A legged robot inspired by spider is needed to access to survivor in search and rescue operations. This paper proposes to control system is based on petri net for six legged spider robot which is used for search and rescue operations. The robotic system is tested by using different walking algorithms. Control of the robot is provided by communication ports on computer. The performance of the robot is calculated entirely, as depending on the movement of six legs on rough terrain. Functional algorithms are created to be moved the robot flexible under difficult conditions such as rough terrain, pit. Also, these algorithms which provide moving of robot at various speeds according to structure of legs are presented. The robot controlled in the project is named as TKSPIDER1 and each leg of it has three servo motors.

- by Tolga Karakurt
- •
- Petri Nets, Legged Robots, Control Systems, Arduino

In this paper, we present a complete description of the hardware design and control architecture of our custom built quadruped robot, called the `Stoch'. Our goal is to realize a robust, modular, and a reliable quadrupedal platform, using which various locomotion behaviors are explored. This platform enables us to explore different research problems in legged locomotion, which use both traditional and learning based techniques. We discuss the merits and limitations of the platform in terms of exploitation of available behaviours, fast rapid prototyping, reproduction and repair. Towards the end, we will demonstrate trotting, bounding behaviors, and preliminary results in turning. In addition, we will also show various gait transitions i.e., trot-to-turn and trot-to-bound behaviors.

- by Shounak Bhattacharya
- •
- Robotics, Computer Science, Mobile Robotics, Legged Robots

The bounding gait for the Platform for Ambulating Wheels (PAW), a new and unique hybrid wheeled-leg system is presented here. Two hypotheses are tested and discussed: first, that the robot’s forward speed can be increased by increasing the leg liftoff angles and, second, that addition of distally-mounted actuated wheels can be used in running gaits such as the bound. Both hypotheses were tested experimentally and found to be valid.

- by James A Smith and +1
- •
- Electrical Engineering, Mechanical Engineering, Artificial Intelligence, Biomedical Engineering
- by Mete Kalyoncu
- •
- Computer Science, Optimization techniques, Legged Robots, Bees Algorithms

Contemporary humanoids are equipped with visual and LiDAR sensors that are effectively utilized for Visual Odometry (VO) and LiDAR Odometry (LO). Unfortunately, such measurements commonly suffer from outliers in a dynamic environment, since frequently it is assumed that only the robot is in motion and the world is static. To this end, robust state estimation schemes are mandatory in order for humanoids to symbiotically co-exist with humans in their daily dynamic environments. In this article, the robust Gaussian Error-State Kalman Filter for humanoid robot locomotion is presented. The introduced method automatically detects and rejects outliers without relying on any prior knowledge on measurement distributions or finely tuned thresholds. Subsequently, the proposed method is quantitatively and qualitatively assessed in realistic conditions with the full-size humanoid robot WALK-MAN v2.0 and the mini-size humanoid robot NAO to demonstrate its accuracy and robustness when outlier VO/LO measurements are present. Finally, in order to reinforce further research endeavours, our implementation is released as an open-source ROS/C++ package.

- by Stylianos Piperakis
- •
- Humanoid Robotics, Legged Robots, State Estimation, Kalman Filter

— In small mobile robots, decreasing the number of actuators is usually desirable to reduce the size and weight of the robot, but it is usually at the expense of the robot's degree of freedom (DOF). This work presents the development and preliminary experimental testing of a novel Legged Piezoelectric Miniature Robot (LPMR) driven only by a single piezoelectric unimorph actuator and yet fully capable of being maneuvered to move forward, turn right, or turn left. The underactuated motion is achieved by exploiting the bending vibration modes disparity of the piezoelectric actuator at different driving frequencies and designing specific positions of the robot's legs to generate a differential-drive-like mechanism. The speed of the robot can be controlled through regulating the magnitude of the applied voltage. The proposed underactuated system is experimentally verified and a preliminary characterization of the LPMR in terms of its forward and turning speed versus applied voltage and payload is investigated and reported.

- by Hassan Hussein Hariri and +1
- •
- Legged Robots, Piezoelectric Actuators
- by Andres Valenzuela
- •
- Humanoid Robotics, Motion Planning, Legged Robots, Legged Locomotion

This paper studies the effect of flexible linear torso on the dynamics of passive quadruped bounding. A reduced-order passive and conservative model with linear flexible torso and springy legs is introduced. The model features extensive spine deformation during high-speed bounding, resembling those observed in a cheetah. Fixed points corresponding to cyclic bounding motions are found and calculated using numerical return map methodologies. Results show that the corresponding robot gaits and the associated performance resemble those of its natural counterparts.

This paper investigates possible sources of variability in the dynamics of legged locomotion, even in itsmost idealized form. The rimless wheel model is a seeminglydeterministic legged dynamic system, popular within the leggedlocomotion community for understanding basic collision dy-namics and energetics during passive phases of walking. Despitethe simplicity of this legged model, however, experimentalmotion capture data recording the passive step-to-step dynamicsof a rimless wheel down a constant-slope terrain actuallydemonstrate signiﬁcant variability, providing strong evidencethat stochasticity is an intrinsic–
and thus unavoidable
–propertyof legged locomotion that should be modeled with care whendesigning reliable walking machines. We present numericalcomparisons of several hypotheses as to the dominant source(s)of this variability: 1) the initial distribution of the angularvelocity, 2) the uneven proﬁle of the leg lengths and 3) thedistribution of the coefﬁcients of friction and restitution acrosscollisions. Our analysis shows that the 3rd hypothesis mostaccurately predicts the noise characteristics observed in ourexperimental data while the 1st hypothesis is also valid forcertain contexts of terrain friction. These ﬁndings suggest thatvariability due to ground contact dynamics, and not simplydue to geometric variations more typically modeled in terrain,is important in determining the stochasticity and resultingstability of walking robots. Although such ground contactvariability might be an expected result in ﬁeld robotics onsigniﬁcantly rough terrain, we again note our experimental dataapplies seemingly deterministic-looking terrains: our resultssuggest that stochastic ground collision models should play animportant role in the analysis and optimization of dynamicperformance and stability in robot walking.

- by Thrishantha Nanayakkara and +1
- •
- Robotics, Nonlinear dynamics, Legged Robots

- by Stylianos Piperakis
- •
- Humanoid Robotics, Legged Robots, State Estimation, Kalman Filter

This paper presents the design and development of a new type of piezoelectric-driven robot, which consists of a piezoelec-tric unimorph actuator integrated as part of the structure of a four-bar linkage to generate locomotion. The unimorph actuator replaces the input link of the four-bar linkage and motion is generated at the coupler link due to the actuator deflection. A dimensional synthesis approach is proposed for the design of four-bar linkage that amplifies the small displacement of the piezoelec-tric actuator at the coupler link. The robot consists of two such piezo-driven four-bar linkages and its gait cycle is described. The robot speed is derived through kinematic modelling and experimentally verified using a fabricated prototype. This result will be important for developing a motion planning control strategy for the robot locomotion, which will be part of future work.

This article presents a summary of applications of chaos and fractals in robotics. Firstly, basic concepts of deterministic chaos and fractals are discussed. Then, fundamental tools of chaos theory used for identifying and quantifying chaotic dynamics will be shared. Principal applications of chaos and fractal structures in robotics research, such as chaotic mobile robots, chaotic behaviour exhibited by mobile robots interacting with the environment, chaotic optimization algorithms, chaotic dynamics in bipedal locomotion and fractal mechanisms in modular robots will be presented. A brief survey is reported and an analysis of the reviewed publications is also presented.

- by Sajid Iqbal
- •
- Engineering, Fractal Geometry, Chaos Theory, Locomotion

Legged robots are being the target of several studies and research. The idea is to develop machines that present characteristics approximate to the ones observed in biological living creatures. However this objective is still relatively far away and the development of prototypes for these studies is expensive and time consuming, which leads to the creation of models that allow the realization of the intended studies in software. These models should include the main characteristics of biological creatures relevant for locomotion studies. Given this, the presented work describes the development of a quadruped robot model in MATLAB/Simmechanics TM. This model is intended to be used in the development of gaits for legged robots based on Central Pattern Generators. With this purpose in mind, the model was developed in a way to accept different gaits by direct introduction of the angular positions of the knee and hip joints. Various parameters of the robot are also easily changed through a configuration file that accompanies the model. This paper presents the model of a robot with flexible body, its legs and its hip and knee joints. The model of a feet-ground interaction was also modelled using a theoretic model described in the literature.

- by Tomas Rodriguez Castro
- •
- Legged Robots, Foot, Force, Kinematic Simulation of Robot

- by Shounak Bhattacharya
- •
- Robotics, Mobile Robotics, Legged Robots, IEEE

Highlights Consideration of the effects of permanent ground deformation and compaction. Development of a controller immune to terrain compliance. No knowledge requirement of ground parameters. Successful tackle of foot slip effects and hard impacts during touchdown. The methodology can be extended to other legged robots such as quadrupeds. Abstract One of the most intriguing research challenges in legged locomotion is robot performance on compliant terrains. The foot-terrain interaction is usually tackled by disregarding some of the effects of ground deformation, like permanent deformation and compaction; however this approach restricts their application to stiff environments. In this work, the foot-terrain interaction is studied, and used in developing a controller immune to terrain compliance. An impact dynamics model is developed, employing a viscoplastic extension a b b

- by Iosif S Paraskevas and +2
- •
- Impact Mechanics, Legged Robots
- by Marco Ceccarelli
- •
- Robotics, Legged Robots, Legged Locomotion

Recent successes of missions such as the MSL and the Rosetta have increased the interest in the robotic exploration of other planets and asteroids. Although most of these missions envisage the use of rovers, legged robots have shown the potential to outperform wheeled vehicles on rough terrains in terms of speed and energy efficiency. In this paper, the x-MP controller presented in recent work, is used to evaluate the performance of a monopod robot under the effect of different gravitational fields and terrain types. The performance of the x-MP controller during regulating the robot motion on rough terrains and for the exploration of different types of planetary environments will be examined using simulations. Additionally using the Cost of Transport index, useful conclusions regarding the performance of legged robots for planetary exploration will be extracted.

- by Iosif S Paraskevas and +2
- •
- Robotics, Impact Mechanics, Legged Robots, Legged Locomotion

Terrain compliance is a critical parameter for the performance of legged locomotion. In this work, a single actuator monopod robot hopping on a rough compliant terrain is considered. Based on our controller for flat compliant terrains, this paper introduces the necessary modifications, which allow the robot to tackle the disturbance of small inclinations. Using the developed method, the robot is examined on its performance to traverse rough terrains, while maintaining the goals of reaching a desired height and forward velocity. As the increased compliance and inclination alter the energy requirements from the controller actuator, the Cost of Transport index for a number of scenarios is studied. The correlation between terrain parameters and the CoT is presented, and useful conclusions, which can aid the understanding of the behavior of legged robots in realistic terrains are extracted.

- by Iosif S Paraskevas and +2
- •
- Robotics, Control Systems Engineering, System Dynamics Modeling, Impact Mechanics
- by Conrad Spiteri and +1
- •
- Image Processing, Legged Robots

This paper presents the concept of variable radius drum mechanisms (VRDMs). A drum, or spool, consists of a spindle with flanges, around which a cable is wound. The cylindrical surface of an ordinary spool has a constant radius. In a variable radius drum (VRD), the radius of the spool varies along its profile. Properties of such devices are discussed, as well as the kinematic analysis and synthesis. The main contribution of the work is the theory of the VRD synthesis problem, rooted in a closed-form analytical solution. In order to highlight the benefits of VRDMs, two applications are presented and analyzed as examples. The first example consists of a mechanism which can support and guide a load along a horizontal linear path. The second example shows a solution to improve the locomotion of a legged robot. Finally, a prototype is made on the basis of the first case scenario and its performance is evaluated and discussed, showing a remarkable accuracy, with a deviation from the nomin...

- by Stefano Seriani
- •
- Engineering, Robotics, Reaction Mechanisms, Theory Of Mechanisms
- by Herbert Tanner
- •
- Engineering, Computer Science, Manufacturing, Legged Robots

- by Dimitrios Kanoulas
- •
- Humanoid Robotics, Legged Robots, State Estimation, Kalman Filter
- by Gauss Lee
- •
- Computer Science, Reinforcement Learning, Locomotion, Morphology

Deep reinforcement learning (deep RL) algorithms leverage the symbolic power of complex controllers by automating it by mapping sensory inputs to low-level actions. Deep RL eliminates the complex robot dynamics with minimal engineering. Deep RL provides high-risk involvement by directly implementing it in real-world scenarios and also high sensitivity towards hyperparameters. Tuning of hyperparameters on a pneumatic quadruped robot becomes very expensive through trial-and-error learning. This paper presents an automated learning control for a pneumatic quadruped robot using sample efficient Deep Q learning, enabling minimal tuning and very few trials to learn the neural network. Long training hours may degrade the pneumatic cylinder due to jerk actions originated through stochastic weights. We applied this method to the pneumatic quadruped robot, which resulted in a hopping gait. In our process, we eliminated the use of a simulator and acquired a stable gait. This approach evolves so that the resultant gait matures more sturdy towards any stochastic changes in the environment. We further show that our algorithm performed very well as compared to programmed gait using robot dynamics.

- by Soofiyan Atar
- •
- Reinforcement Learning, Machine Learning, Neural Networks, Legged Robots

The bounding gait for the Platform for Ambulating Wheels (PAW), a new and unique hybrid wheeled—leg system is presented. Two hypotheses are tested and discussed: first, that the robot’s forward speed can be increased by increasing the leg liftoff angles and, second, that the addition of distally mounted actuated wheels can be used in running gaits such as the bound. Both hypotheses were tested experimentally and found to be valid.