Mobile legged robots have the potential to restructure many aspects of our lives in the near future. Whether for applications in household care, entertainment, or disaster response, these systems depend on high-performance actuators to improve their basic capabilities. The work presented here focuses on developing new high-performance actuators, specifically series elastic actuators, to address this need. We adopt a system-wide optimization approach, dealing with factors which influence performance at the levels of mechanical design, electrical system design, and control. Using this approach and based on a set of performance metrics, we produce an actuator, the UT-SEA, which achieves leading empirical results in terms of power-to-weight, force control, size, and system efficiency. We also develop general high-performance control techniques for both force- and position-controlled actuators, some of which were adopted for use on NASA-JSC's Valkyrie Humanoid robot and were used during DARPA's DRC Trials 2013 robotics competition. / text

This thesis designed and realized a two-degree of freedom wrist and two finger manipulator that completes the six-degree of freedom Sensitive Manipulation Platform, the arm of which was previously developed. This platform extends the previous research in the field of robotics by covering not only the end effector with deformable tactile sensors, but also the links of the arm. Having tactile sensors on the arm will improve the dynamic model of the system during contact with its environment and will allow research in contact navigation to be explored. This type of research is intended for developing algorithms for exploring dynamic environments. Unlike traditional robots that focus on collision avoidance, this platform is designed to seek out contact and use it to gather important information about its surroundings. This small desktop platform was designed to have similar proportions and properties to a small human arm. These properties include compliant joints and tactile sensitivity along the lengths of the arms. The primary applications for the completed platform will be research in contact navigation and manipulation in dynamic environments. However, there are countless potential applications for a compliant arm with increased tactile feedback, including prosthetics and domestic robotics. This thesis covers the details behind the design, analysis, and evaluation of the two degrees of the Wrist and two two-link fingers, with particular attention being given to the integration of series elastics actuators, the decoupling of the fingers from the wrist, and the incorporation of tactile sensors in both the forearm motor module and fingers.

Grasp stability was studied and researched upon by various research groups, but mainly focused on robotic grippers by devising conditions for a stable grasp. Maintaining grasp stability is important so as to reduce the chances of the object slipping and dropping. But there was little focus on the grasp stability of robotic exoskeleton gloves and most of the research was focused on mechanical design. A robotic exoskeleton glove was developed as well as novel methods to improve the grasp stability. The exoskeleton glove developed is intended for patients who have suffered paralysis of the hand due to stroke or other factors. The robotic glove aids them in grasping objects as part of daily life activities. The glove is constructed with rigidly coupled 4-bar linkages attached to the finger tips. Each linkage mechanism has 1- Degree of Freedom (DOF) and is actuated by a linear Series Elastic Actuator (SEA). Two methods were developed to satisfy two of the conditions required for a stable grasp. These include deformation prevention of soft objects, and maintaining force and moment equilibrium of the objects being grasped. Simulations were performed to validate the performance of the algorithms. A battery of experiments was performed on the integrated prototype in order to validate the performance of the algorithms developed. / Master of Science / An exoskeleton glove is robotic device that can aid people who suffer from paralysis of their hands caused by a stroke or other factors with the primary goal of allowing them to regain the basic ability of grasping objects and thereby improving their quality of life. The exoskeleton glove developed in this research is focused on objects grasping assistance rather than for rehabilitation purposes. Since the exoskeleton glove lacks conscious senses like a human hand typically possesses, it may not be able to apply sufficient grasping force or may apply excessive force than required irrespective of the object being grasped. In order to ensure that the exoskeleton glove applies the proper amount of force, two novel methods were developed which help improve the overall grasping performance of the robotic glove. These methods use sensors that enable the glove to react to the force interaction changes that exists between the hand and the object being grasped through the exoskeleton glove. The first method detects any deformation that may occur while grasping a soft object and applies lesser force accordingly to prevent further damage to the object. The second method uses motion sensor to detect any movement by the user while grasping the object and applies corrective forces so that the object doesn’t slip from the hand. A prototype was designed and integrated and the two methods were tested on the prototype to validate them.

This paper discusses the design and implementation of a high performance, custom control solution for series elastic actuators (SEA) in a parallel or serial configuration. In many modern robotics applications, controlling actuator output force accurately and with high bandwidth is extremely important. The series elastic actuator has become popular in applications which require precise force control, however currently not many commercial options exist. Commonly, these actuators are custom designed and use electric motors, however most off-the-shelf electric motor drives are not designed for this specific application. In this paper, the hardware and software architecture of a control device designed specifically for force controlled series elastic actuators is described, along with test results on a novel SEA design. / Master of Science

We present work to transfer decentralized neuromuscular control strategies of human locomotion to powered segmented robotic legs. State-of-the-art robotic locomotion control approaches, like centralized planning and tracking in fully robotic systems and predefined motion pattern replay in prosthetic systems, do not enable the dynamism and reactiveness of able-bodied humans. Animals largely realize dexterous segmented leg performance with leg-encoded biomechanics and local feedback controls that bypass central processing. A decentralized neuromuscular controller was recently developed that enables robust locomotion for a simulated multi-segmented planar humanoid. A portion of this controller was used in an active ankle-foot prosthesis to modulate ankle torque during stance, enabling level and inclined ground walking. While these results suggest that the neuromuscular controller is a promising alternative control method for both fully robotic systems and powered prostheses, it is unclear if the controller can be transferred to multi-segmented robotic legs. The goal of this thesis is to investigate the feasibility of controlling a multi-segmented robotic leg with the proposed neuromuscular control approach, which may enable robots and powered prostheses to react to locomotion disturbances dynamically and in a human-like way. Specifically, work in this thesis investigates two hypotheses. Hypothesis one posits that the proposed decentralized swing-leg controllers enable more robust foot placements into ground targets than state-of-the-art impedance controls. Hypothesis two posits that neuromuscular swing-leg control enables more human-like motion than state-of-the-art impedance control. To transfer neuromuscular controls to powered segmented robotic legs, we use a model-based design approach. The initial transfer is focused on neuromuscular swing-leg controls, important for maintaining dynamic stability of both fully robotic systems and powered prostheses in the presence of unexpected locomotion disturbances, such as trips and pushes. We first present the design of RNL, a three segment, cable-driven, antagonistically actuated robotic leg with joint compliance. The robot’s size, weight, and actuation capabilities correspond to dynamically scaled human values. Next, a highfidelity simulation of the robot is created to investigate the feasibility of transferring neuromuscular controls, pre-tune hardware gains via optimization, and serve as a benchmark for hardware experiments. An idealized version of the swing-leg controller with mono-articular actuation, as well as the neuromuscular interpretation of this controller with multi-articular actuation is then transferred to RNL and evaluated with foot placement experiments. The results suggest that the proposed swing-leg controllers can accurately regulate foot placement of robotic legs during undisturbed and disturbed motions. Compared to impedance control, the proposed controls achieve foot placements over a range of ground targets with a single set of gains, which make them attractive candidates for regulating the motion of legged robots and prostheses in the real-world. Furthermore, the ankle trajectory traced out by the robot under neuromuscular control is more human-like than the trajectories traced out under the proposed idealized control and impedance control. In parallel to this control transfer, a synthesis method for creating compact nonlinear springs with user-defined torque-deflection profiles is presented to explore methods for improving RNL’s series elastic actuators. The springs use rubber as their elastic element, which, while enabling a compact spring design, introduce viscoelastic behavior in the spring that needs to be accounted for with additional control. To accurately estimate force developed in the rubber, an empirically characterized constitutive rubber model is developed and integrated into the series elastic actuator controller used by the RNL test platforms. Benchtop experiments show that in conjunction with an observer, the nonlinear spring prototype achieves desired behavior at actuation frequencies up to 2 Hz, after which spring behavior degrades due to rubber hysteresis. These results show that while the presented methodology is capable of realizing compact nonlinear springs, careful rubber selection that mitigates viscoelastic behavior is necessary during the spring design process.

The motor control of anthropomorphic robotic systems is a challenging computational task mainly because of the high levels of redundancies such systems exhibit. Optimality principles provide a general strategy to resolve such redundancies in a task driven fashion. In particular closed loop optimisation, i.e., optimal feedback control (OFC), has served as a successful motor control model as it unifies important concepts such as costs, noise, sensory feedback and internal models into a coherent mathematical framework. Realising OFC on realistic anthropomorphic systems however is non-trivial: Firstly, such systems have typically large dimensionality and nonlinear dynamics, in which case the optimisation problem becomes computationally intractable. Approximative methods, like the iterative linear quadratic gaussian (ILQG), have been proposed to avoid this, however the transfer of solutions from idealised simulations to real hardware systems has proved to be challenging. Secondly, OFC relies on an accurate description of the system dynamics, which for many realistic control systems may be unknown, difficult to estimate, or subject to frequent systematic changes. Thirdly, many (especially biologically inspired) systems suffer from significant state or control dependent sources of noise, which are difficult to model in a generally valid fashion. This thesis addresses these issues with the aim to realise efficient OFC for anthropomorphic manipulators. First we investigate the implementation of OFC laws on anthropomorphic hardware. Using ILQG we optimally control a high-dimensional anthropomorphic manipulator without having to specify an explicit inverse kinematics, inverse dynamics or feedback control law. We achieve this by introducing a novel cost function that accounts for the physical constraints of the robot and a dynamics formulation that resolves discontinuities in the dynamics. The experimental hardware results reveal the benefits of OFC over traditional (open loop) optimal controllers in terms of energy efficiency and compliance, properties that are crucial for the control of modern anthropomorphic manipulators. We then propose a new framework of OFC with learned dynamics (OFC-LD) that, unlike classic approaches, does not rely on analytic dynamics functions but rather updates the internal dynamics model continuously from sensorimotor plant feedback. We demonstrate how this approach can compensate for unknown dynamics and for complex dynamic perturbations in an online fashion. A specific advantage of a learned dynamics model is that it contains the stochastic information (i.e., noise) from the plant data, which corresponds to the uncertainty in the system. Consequently one can exploit this information within OFC-LD in order to produce control laws that minimise the uncertainty in the system. In the domain of antagonistically actuated systems this approach leads to improved motor performance, which is achieved by co-contracting antagonistic actuators in order to reduce the negative effects of the noise. Most importantly the shape and source of the noise is unknown a priory and is solely learned from plant data. The model is successfully tested on an antagonistic series elastic actuator (SEA) that we have built for this purpose. The proposed OFC-LD model is not only applicable to robotic systems but also proves to be very useful in the modelling of biological motor control phenomena and we show how our model can be used to predict a wide range of human impedance control patterns during both, stationary and adaptation tasks.

A novel hip exoskeleton with six degrees of freedom (DoF) was developed, and multiple prototypes of this product were created in this thesis. The device was an upper level of the 12-DoF lower-body exoskeleton project, which was known as the Orthotic Lower-body Locomotion Exoskeleton (OLL-E). The hip exoskeleton had three motions per leg, which were roll, yaw, and pitch. Currently, the sufferers of hemiplegia and paraplegia can be addressed by using a wheelchair or operating an exoskeleton with aids for balancing. The motivation of the exoskeleton project was to allow paraplegic patients to walk without using aids such as a walker or crutches. In mechanical design, the hip exoskeleton was developed to mimic the behavior of a healthy person closely. The hip exoskeleton will be fully powered by a custom linear actuator for each joint. To date, there are no exoskeleton products that are designed to have all of the hip joints powered. Thus, packaging of actuators was also involved in the mechanical design of the hip exoskeleton. As a result, the output torque and speed for the roll joint and yaw joint were calculated. Each hip joint was structurally designed with properly selected bearings, encoder, and hard stops. Their range of motions met desired requirements. In addition, a backpack assembly was designed for mounting the hardware, such as cooling pumps, radiators, and batteries. In the verification part, finite element analysis (FEA) was conducted to show the robustness of the structural design. For fit testing, three wearable prototypes were produced to verify design choices. As a result, the weight of the current hip exoskeleton was measured as 32.1 kg. / Master of Science

Embedded systems for humanoid robots are required to be reliable, low in cost, scalable and robust. Most of the applications related to humanoid robots require efficient force control of Series Elastic Actuators (SEA). These control loops often introduce precise timing requirements due to the safety critical nature of the underlying hardware. Also the motor controller needs to run fast and interface with several sensors. The commercially available motor controllers generally do not satisfy all the requirements of speed, reliability, ease of use and small size. This work presents a custom motor controller, which can be used for real time force control of SEA on humanoid robots and exoskeletons. Emphasis has been laid on designing a system which is scalable, easy to use and robust. The hardware and software architecture for control has been presented along with the results obtained on a novel Series Elastic Actuator based humanoid robot THOR. / Master of Science

This study focuses on the design and implementation of a humanoid robot for disaster response. In particular, this thesis investigates the lower body design in detail with the upper body discussed at a higher level. The Tactical Hazardous Operations Robot (THOR) was designed to compete in the DARPA Robotics Challenge where it needs to complete tasks based on first-responder operations. These tasks, ranging from traversing rough terrain through driving a utility vehicle, suggest a versatile platform in a human sized form factor. A physical experiment of the proposed tasks generated a set of joint range of motions (RoM). Desired limb lengths were determined by comparing existing robots, the test subject in the experiment of proposed tasks, and an average human. Simulations using the desired RoM and limb lengths were used to calculate baseline joint torques. Based on the generated design constraints, THOR is a 34 degree of freedom humanoid that stands 1.78 [m] tall and weighs 65 [kg]. The 12 lower body joints are driven by series elastic linear actuators with multiple joints actuated in parallel. The parallel actuation mimics the human body, where multiple muscles pull on the same joint cooperatively. The legs retain high joint torques throughout their large RoM with some joints achieving torques as high as 289 [Nm]. The upper body uses traditional rotary actuators to drive the waist, arms, and head. The proprioceptive sensor selection was influenced by past experience on humanoid platforms, and perception sensors were selected to match the competition. / Master of Science

A novel rotary series elastic actuator (RSEA) with a two-mode, or bimodal, elastic element was designed and tested. This device was developed to eliminate the compromise between human safety and robot performance. Rigid actuators can be dangerous to humans within a robot's workspace due to impacts or pinning scenarios. To increase safety, elastic elements can soften impacts and allow for escape should pinning occur. However, adding elasticity increases the complexity of the system, lowers the bandwidth, and can make control of the actuator more difficult. To get the best of both types of actuators, a bimodal clutch was designed to switch between rigid actuation for performance and elastic actuation for human safety. The actuator consisted of two main parts, a rigid rotary actuator using a harmonic gearhead and a drum brake designed to act as a clutch. The 200 W rotary actuator provides 54.7 Nm of torque with a maximum speed of 41.4 rpm. The measured efficiency was 0.797 due to a timing belt speed reduction that was then speed reduced with a harmonic gearhead. The clutch was a drum brake actuated with a pantograph linkage and ACME lead screw. This configuration produced 11 Nm of holding torque experimentally but was theoretically shown to produce up to 51.4 Nm with larger motors. The elastic element was designed using finite element analysis (FEA) and tested experimentally to find a measured stiffness of 290 Nm/rad. / Master of Science