Humanoid robots possess the ability to perform complex tasks in challenging environments. However, they require a model of the surroundings in a representation that is sufficient enough for downstream tasks such as footstep planning. The maps generated by existing mapping algorithms are either sparse, insufficient for footstep planning, memory intensive, or too slow for dynamic humanoid behaviors. In this work, we develop a mapping algorithm that combines planar region measurements along with kinematic-inertial state estimates to build a dense but efficient map of bounded planar surfaces. We present novel algorithms for plane feature matching, tracking and registration for mapping within a factor graph framework. The generated map is not only memory efficient, but also offers higher reliability and speed in bipedal footstep planning, than was possible earlier. The complete algorithm is also demonstrated using a full-scale humanoid robot, Nadia, walking over both flat ground and rough terrain utilizing the generated terrain map.

Complex multibody legged robots can have complex rotational control challenges. In this paper, we propose a concise way to understand and formulate a whole-body orientation that (i) depends on system configuration only and not a history of motion, (ii) can be representative of the orientation of the entire system while not being attached to any specific link, and (iii) has a rate of change that approximates total system angular momentum. We relate this orientation coordinate to past work, and discuss and demonstrate, including on hardware, several different uses for it.

Teleoperation of humanoid robots enables the integration of the cognitive skills and domain expertise of humans with the physical capabilities of humanoid robots. The operational versatility of humanoid robots makes them the ideal platform for a wide range of applications when teleoperating in a remote environment. However, the complexity of humanoid robots imposes challenges for teleoperation, particularly in unstructured dynamic environments with limited communication. Many advancements have been achieved in the last decades in this area, but a comprehensive overview is still missing. This survey article gives an extensive overview of humanoid robot teleoperation, presenting the general architecture of a teleoperation system and analyzing the different components. We also discuss different aspects of the topic, including technological and methodological advances, as well as potential applications.

Obtaining highly dynamic motions in robots is a difficult task. In recent years, sophistication in mechanical design, improved algorithms, and high computational power allows new robots to perform natural gaits and dynamic motions such as backflips. Offline optimization is often necessary to obtain good performance in those difficult motions. However, when an athlete does a backflip, he will adapt “online” to any change, and that is shown in the robustness of the movements. One of the biggest challenges in robotics is to perform those movements using online optimization with the dynamics of the robot. Here, we present an approach to deal with complicated tasks using online optimization. We obtain 90° rotational jumps and jumps over sloped terrain in the Mini-Cheetah hardware, and online-optimized backflips, sideflips, and frontflips in a real-time physical simulator with full-body dynamics.

Awareness of the environment is essential for mobile robots. Perception for legged robots requires high levels of reliability and accuracy in order to walk stably in the types of complex, cluttered environments we are interested in. In this paper, we present a usable environmental perception algorithm designed to detect steppable areas and obstacles for the autonomous generation of desired footholds for legged robots. To produce an efficient representation of the environment, the proposed perception algorithm is desired to cluster point cloud data to planar regions composed of convex polygons. We describe in this paper the end-to-end pipeline from data collection to generation of the regions, where we first compose an octree in order to create a more efficient data representation. We then group the leaves in the tree using a nearest neighbor search into a planar region, which is composed of the concave hull of points that is decomposed into convex polygons. We present a variety of environments, and illustrate the usability of this approach by the Atlas humanoid robots walking over rough terrain. We also discuss various challenges we faced and insights we gained in the development of this approach.

Performing large step-ups is a challenging task for a humanoid robot. It requires the robot to perform motions at the limit of its reachable workspace while straining to move its body upon the obstacle. This paper presents a non-linear trajectory optimization method for generating step-up motions. We adopt a simplified model of the centroidal dynamics to generate feasible Center of Mass trajectories aimed at reducing the torques required for the step-up motion. The activation and deactivation of contacts at both feet are considered explicitly. The output of the planner is a Center of Mass trajectory plus an optimal duration for each walking phase. These desired values are stabilized by a whole-body controller that determines a set of desired joint torques. We experimentally demonstrate that by using trajectory optimization techniques, the maximum torque required to the full-size humanoid robot Atlas can be reduced up to 20% when performing a step-up motion.

Legged humanoid robots promise revolutionary mobility and effectiveness in environments built for humans. However, inefficient use of energy significantly limits their practical adoption. The humanoid biped walking anthropomorphic novelly-driven efficient robot for emergency response (WANDERER) achieves versatile, efficient mobility, and high endurance via novel drive-trains and passive joint mechanisms. Results of a test in which WANDERER walked for more than 4 h and covered 2.8 km on a treadmill, are presented. Results of laboratory experiments showing even more efficient walking are also presented and analyzed in this article. WANDERER's energetic performance and endurance are believed to exceed the prior literature in human-scale humanoid robots. This article describes WANDERER, the analytical methods and innovations that enable its design, and system-level energy efficiency results.

We present an approach for achieving a natural, efficient gait on bipedal robots using straightened legs and toe-off. Our algorithm avoids complex height planning by allowing a whole-body controller to determine the straightest possible leg configuration at run-time. The controller solutions are biased towards a straight leg configuration by projecting leg joint angle objectives into the null-space of the other quadratic program motion objectives. To allow the legs to remain straight throughout the gait, toe-off was utilized to increase the kinematic reachability of the legs. The toe-off motion is achieved through underconstraining the foot position, allowing it to emerge naturally. We applied this approach of under-specifying the motion objectives to the Atlas humanoid, allowing it to walk over a variety of terrain. We present both experimental and simulation results and discuss performance limitations and potential improvements.

When walking at high speeds, the swing legs of robots produce a non-negligible angular momentum rate. To accommodate this, we provide a reference trajectory generator for bipedal walking that incorporates predicted centroidal angular momentum at the planning stage. This can be done efficiently as the Centroidal Moment Pivot (CMP), Instantaneous Capture Point (ICP) and the center of mass (CoM) all have closed-form trajectory solutions due to their linear dynamics. This is then used to produce smooth, continuous trajectories. We furthermore provide a lightweight model to estimate angular momentum as induced during leg swing of the gait cycle. Our proposed trajectory generator is tested thoroughly in simulation and has been shown to successfully operate on the real hardware.

While humans are highly capable of recovering from external disturbances and uncertainties that result in large tracking errors, humanoid robots have yet to reliably mimic this level of robustness. Essential to this is the ability to combine traditional “ankle strategy” balancing with step timing and location adjustment techniques. In doing so, the robot is able to step quickly to the necessary location to continue walking. In this work, we present both a new swing speed up algorithm to adjust the step timing, allowing the robot to set the foot down more quickly to recover from errors in the direction of the current capture point dynamics, and a new algorithm to adjust the desired footstep, expanding the base of support to utilize the center of pressure (CoP)-based ankle strategy for balance. We then utilize the desired centroidal moment pivot (CMP) to calculate the momentum rate of change for our inverse-dynamics based whole-body controller. We present simulation and experimental results using this work, and discuss performance limitations and potential improvements.

In this paper we introduce STEPPR (Sandia Transmission-Efficient Prototype Promoting Research), a bipedal robot designed to explore efficient bipedal walking. The initial iteration of this robot achieves efficient motions through powerful electromagnetic actuators and highly back-drivable synthetic rope transmissions. We show how the addition of parallel elastic elements at select joints is predicted to provide substantial energetic benefits: reducing cost of transport by 30 to 50 percent. Two joints in particular, hip roll and ankle pitch, reduce dissipated power over three very different gait types: human walking, human-like robot walking, and crouched robot walking. Joint springs based on this analysis are tested and validated experimentally. Finally, this paper concludes with the design of two unique parallel spring mechanisms to be added to the current STEPPR robot in order to provide improved locomotive efficiency.

This paper works with the concept of Divergent Component of Motion (DCM), also called `(instantaneous) Capture Point'. We present two real-time DCM trajectory generators for uneven (three-dimensional) ground surfaces, which lead to continuous leg (and corresponding ground reaction) force profiles and facilitate the use of toe-off motion during double support. Thus, the resulting DCM trajectories are well suited for real-world robots and allow for increased step length and step height. The performance of the proposed methods was tested in numerous simulations and experiments on IHMC's Atlas robot and DLR's humanoid robot TORO.

Bipedal robots are currently either slow, energetically inefficient and/or require a lot of control to maintain their stability. This paper introduces the FastRunner, a bipedal robot based on a new leg architecture. Simulation results of a Planar FastRunner demonstrate that legged robots can run fast, be energy efficient and inherently stable. The simulated FastRunner has a cost of transport of 1.4 and requires only a local feedback of the hip position to reach 35.4 kph from stop in simulation.

This paper presents work by researchers at The Florida Institute for Human and Machine Cognition (IHMC) on the DARPA funded Learning Locomotion program. This program started in 2005 and finished in 2009. The goal of the Learning Locomotion program was to develop control algorithms for autonomous traversal of large, irregular obstacles by an unmanned quadrupedal robot. The program was designed to allow multiple teams to compete for the fastest speeds across the terrain boards using identical hardware. Terrain sensing and localization issues were eliminated by providing a high resolution height map of the terrain as well as real time pose information of the robot on the terrain. The objective for each trial was to reach the goal position as fast as possible.

The IHMC Mobility Assist Exoskeleton is a robotic suit that a user can wear for strength augmentation or gait generation. This first generation exoskeleton prototype focuses on providing walking assistance to persons with lower extremity paralysis. The main goal is to successfully enable a person that cannot walk without assistance to walk in a straight line a distance of 15 feet. When in disable assist mode this prototype will rely on the user to provide balance control, and thus an external means for balancing will be required, such as crutches or a walker. Power and control is off board and supplied to the Exoskeleton by means of a tether. Rotary Series Elastic Actuators (RSEAs), which have high force fidelity and low impedance were designed to power the joints. This paper describes the design, test results, future work and potential applications of the exoskeleton.

Current fully autonomous robots are unable to navigate effectively in visually complex environments due to limitations in sensing and cognition. Full teleoperation using current interfaces is difficult and the operator often makes navigation mistakes due to lack of operating environment information and a limited field of view. We present a novel method for combining the sensing and cognition of a robot with that of a human. Our collaborative approach is different from most in that we address bi-directional considerations. It provides the human a mechanism to supplement the robot's capabilities in a new and unique way and provides novel forms of feedback from the robot to enhance the human's understanding of the current state of the system and its intentions.

This video highlights work to date on the Yobotics-IHMC lower body humanoid robot. The robot is a twelve degree-of-freedom robot with force controllable series elastic actuators at each degree of freedom. Control algorithms utilize virtual model control, and foot placement is determined using capture regions. The robot can recover from moderate disturbances and walk on flat ground. Ongoing work is focused on improving robustness to disturbances, walking more quickly and efficiently, and walking over rough terrain.

Researchers at IHMC and Honda Research Institute are developing techniques for learning capture points for bipedal push recovery. A capture point is a point on the ground where a biped can step to in order to stop. Humans are very adept at stepping to capture points, while most bipedal robots cannot recover from significant pushes. To calculate approximate capture point locations, we use the linear inverted pendulum model introduced by Kajita and Tani. For a point mass biped walking at a constant height, this model exactly predicts the capture point. However, for a distributed mass biped, it is only an approximation. In order to better predict capture points, we learn a correction function to the linear inverted pendulum model. We used two learning methods, one online and one offline, to improve capture point prediction. In the offline learning method, the robot is pushed multiple times with a given force magnitude and direction. In the online learning technique, we use a radial basis function to represent the learned offsets from the capture point predicted by the linear inverted pendulum model.

We present an analysis of a point mass, point foot, planar inverted pendulum model for bipedal walking. Using this model, we derive expressions for a conserved quantity, the "orbital energy", given a smooth center of mass trajectory. Given a closed form center of mass trajectory, the equation for the orbital energy is a closed form expression except for an integral term, which we show to be the first moment of area under the center of mass path. Hence, given a center of mass trajectory, it is straightforward and computationally simple to compute phase portraits for the system. In fact, for many classes of trajectories, such as those in which height is a polynomial function of center of mass horizontal displacement, the orbital energy can be solved in closed form. Given expressions for the orbital energy, we can compute where the foot should be placed or how the center of mass trajectory should be modified in order to achieve a desired velocity on the next step. We demonstrate our results using a planar biped simulation with light legs and point mass body. We parameterize the center of mass trajectory with a fifth order polynomial function. We demonstrate how the parameters of this polynomial and step length can be changed in order to achieve a desired next step velocity.

We present a controller for a quadrupedal robot statically walking on known rough terrain. The controller has both deliberative and reactive components for task specific control issues, such as impassable terrain and unmodeled foot slippage. The controller architecture supports multiple gaits, and we present both a stable omnidirectional gait and a faster directional gait. The robot successfully negotiates obstacles up to 7.5 cm (ap40% leg length) tall and navigates over rocky terrain.

We present several biologically inspired concept designs and feasibility analyses for underwater swimming exoskeletons. These designs are biologically inspired, based on observations of dolphins, sea turtles, and penguins. Biologically inspired designs have the advantages of stealth, maneuverability, and a natural interface when compared to propeller driven underwater propulsion devices. We present a lower body concept, based on dolphin locomotion and an upper body concept based on sea turtle and penguin locomotion. The dolphin based concept has the advantages of using the wearer's most powerful muscle groups, of being the most natural to swim with, and of leaving the user's hands free for other tasks. The sea-turtle based concept has the advantage of novelty and may have appeal as a recreational device. We predict that with actuation and energy storage components available today, an exoskeleton that produces a cruising speed of over 1 m/s and a top speed of over 1.5 m/s is feasible. We estimate that cruising at 1 m/s can be achieved with less than 504 Watts power consumption, which translates into 2.4 kg of off-the-shelf silver-zinc batteries per hour of operation.

Exoskeletons that enhance human strength, endurance, and speed while being transparent to the wearer are feasible. In order to be transparent, the exoskeleton must determine the user's intent, apply forces when and where appropriate, and present low impedance to the wearer. We present a one degree of freedom exoskeleton called the RoboKnee which achieves a high level of transparency. User intent is determined through the knee joint angle and ground reaction forces. Torque is applied across the knee in order to allow the user's quadriceps muscles to relax. Low impedance is achieved through the use of series elastic actuators. The RoboKnee allows the wearer to climb stairs and perform deep knee bends while carrying a significant load in a backpack. The device provides most of the energy required to work against gravity while the user stays in control, deciding when and where to walk, as well as providing balance and control. Videos, photographs, and more information about the RoboKnee can be found at http://www.yobotics.com.

Simple intuitive control strategies can be used to compel bipedal robots to walk over sloped terrain. We describe an algorithm for walking dynamically and steadily over sloped terrain with unknown slope gradients and transition locations. The algorithm is developed based on geometric considerations. The overall algorithm is very simple and does not require the biped to have an extensive sensory system for walking over moderate slopes. The ground is detected blindly using only foot contact switches. Using a few simple strategies, we have compelled a simulated 7-link planar biped to walk up and down slopes and over rolling terrain.

We present a stable adaptive control approach for a bipedal walking robot. This approach utilizes a self-organizing CMAC neural network mechanism which has a fast training rate, high approximation accuracy and significant reduction in space complexity. In order to apply this control approach to a bipedal walking robot, a Cartesian virtual dynamics space is introduced based on the virtual model control concept. The adaptive CMAC neural network control approach identifies the unmodelled dynamics of the bipedal robot and ensures asymptotic system stability in a Lyapunov sense. It can also better accommodate unexpected external disturbances, enhancing the control robustness of the bipedal robot. The CMAC neural network structure, its training algorithm, and bipedal locomotion control are described. The simulation results for a walking robot are presented.

Bipedal robots are difficult to analyze mathematically. However, successful control strategies can be discovered using simple physical intuition and can be described in simple terms. Five things have to happen for a planar bipedal robot to walk. Height has to be stabilized. Pitch has to be stabilized. Speed has to be stabilized. The swing leg has to move so that the feet are in locations which allow for the stability of height, pitch, and speed. Finally, transitions from support leg to support leg must occur at appropriate times. If these five objectives are achieved, the robot will walk. A number of different intuitive control strategies can be used to achieve each of these five objectives. Further, each strategy can be implemented in a variety of ways. We present several strategies for each objective which we have implemented on a bipedal walking robot. Using these simple intuitive strategies, we have compelled a seven link planar bipedal robot, called Spring Flamingo, to walk. The robot walks both slowly and quickly, walks over moderate obstacles, starts, and stops.

The robustness of biped walking can be enhanced by the use of adaptive control and learning. The paper describes one such approach, radial basis function (RBF) neural network adaptive control (NNAC). The adaptive control mechanism is designed in a virtual space utilizing the virtual model control paradigm. The neural network is parameterized and trained in an unsupervised learning mode. There are two advantages to this approach. First, the NNAC can identify the unmodelled dynamics of the robot and ensure asymptotic system stability in a Lyapunov sense. Second, the controller can better accommodate unexpected external disturbances. The system's design is described and simulation results are presented.

The transformation from high level task specification to low level motion control is a fundamental issue in sensorimotor control in animals and robots. This paper describes a control scheme called virtual model control that addresses this issue. Virtual model control is a motion control language that uses simulations of imagined mechanical components to create forces, which are applied through real joint torques, thereby creating the illusion that the virtual components are connected to the robot. Due to the intuitive nature of this technique, designing a virtual model controller requires the same skills as designing the mechanism itself. A high level control system can be cascaded with the low level virtual model controller to modulate the parameters of the virtual mechanisms. Discrete commands from the high level controller would then result in fluid motion. Virtual model control has been applied to a physical bipedal walking robot. A simple algorithm utilizing a simple set of virtual components has successfully compelled the robot to walk continuously over level terrain.

Robots typically have an individual actuator at each joint which can result in a nonintuitive and difficult control problem. In this paper we present a control method in which the real joint actuators are used to mimic virtual actuators which can be more intuitive and hence make the control problem more straightforward. Our virtual actuator control method requires a solution to the force distribution problem when applied to parallel mechanisms. An extension of Gardner's partitioned actuator set control method (1991) is presented. This extended method allows for dealing with constrained degrees of freedom in which the torque cannot be specified but can be measured. A simulated hexapod robot was developed to test the proposed control method. The virtual actuators allowed textbook control solutions to be used in controlling this highly nonlinear, parallel mechanism. Using a simple linear control law, the robot walked while simultaneously balancing a pendulum and tracking an object.