In recent years, the deployment of ground-based mobile robots has gained more and more interest in various domains. In contrast to other types of mobile robots, legged robots can traverse irregular terrains, climb stairs, and step over obstacles. However, these unique properties intensify the energy demand and require highly advanced perception methods, actuator designs, and motion control algorithms. The most significant challenges in legged robotics lie in robustness, energy efficiency, and agility.

In recent years, the integration of an articulated torso or active spine, inspired by the body motion of high-performance mammals like the cheetah, has shown promising results. Various studies observed higher maximum velocities and lower energy consumption compared to a rigid torso. However, in these studies, the compliant systems were typically controlled using basic control strategies. In recent years, the development of highly dynamic model-based motion optimization strategies has greatly enhanced the overall performance of various legged robots. Therefore, a model-based motion optimization scheme is developed specifically for articulated quadruped robots. This scheme fully exploits the additional degrees of freedom of the torso to enhance the dynamic performance of the legged robot further.

In this thesis, we introduce a novel approach aimed at enhancing the jumping and landing capabilities of quadruped robots. Our method integrates both model-based and model-free strategies and features a behavioral cloning framework designed to reduce computational delays often encountered in trajectory optimization.

Initially, we build upon an existing framework for quadruped jumps, where we refine the trajectory optimization (TO) algorithm and introduce a new Variable Impedance Control (VIC). The VIC is specifically developed to facilitate softer landings. This improved system was then utilized to generate a comprehensive synthetic dataset, including 11,000 samples that cover a diverse range of jumping scenarios. This dataset served as the foundation for training a neural network. The primary objective of the network is to emulate the performance of the model-based approach. Structurally, the network is designed to process the robot's current state as input and generate the corresponding control actions for its 12 motors as output.

The most significant achievement of this research is the neural network's ability to closely replicate the outcomes of the model-based solution. Notably, it ensures more compliant behavior and lower stress on the motors during the landing phase than an MPC. The neural network demonstrates a 97.4% success rate. This high level of performance underscores its potential for on-the-fly application in robotic systems. The effectiveness of our method is further validated through a series of simulations and practical tests conducted on a Go1 quadruped robot.

In the field of fluid mechanics, there has been a significant shift towards the integration of machine and deep learning techniques to address challenges in reduced-order modeling, flow feature analysis, and control, especially within the realm of active flow control (AFC) for objectives such as lift optimization and drag reduction. Deep learning has taken a central role in advancing state-of-the-art AFC methods by creating data-driven models that mitigate the computational demands of conventional Computational Fluid Dynamics (CFD) simulations, enabling real-time fluid control. Despite the predominance of models trained offline and focused on simple scenarios like laminar flow around bluff bodies, the utility of sophisticated learning methods in AFC has remained largely unexplored.

This research introduces a novel benchmark in fluid dynamics—a soft robotic tentacle actuator—to evaluate the effectiveness of deep learning architectures in complex flow control situations. Through the comparison of online and offline learning frameworks for predicting system behavior, the study elucidates the strengths and limitations of deep learning networks in AFC. The findings underscore the constraints faced by deep learning architectures when dealing with aperiodic motions and demonstrate the significant benefits of adopting an online learning approach over offline training methods, thereby highlighting the advantages of adaptive learning strategies in complex AFC scenarios. The online learning framework displays more stability and increased quality of forecasts at larger time horizons.

The loss of an upper limb significantly impacts an individual's life, directly affecting their ability to perform activities of daily living. Prosthetic devices play a crucial role in aiding amputees' rehabilitation in society, primarily driven by advancements in externally powered prostheses. Although prosthetic devices on the market offer excellent functionality, they come with a high price tag and use complex control algorithms.
Over recent years, a noticeable shift towards simplifying prosthetic devices has emerged, often facilitated by soft robotic principles. For example, the adaptive synergy approach has led to devices that are highly adaptable to their environment with a reduced degree of actuation (DoA), thereby improving functionality at low complexity. This thesis explores a novel research direction by combining the concept of adaptive synergy actuation with additional, parallel actuation of individual fingers. The main goal of this research is to assess the viability of using such a parallel adaptive synergy actuation structure for prosthetic hands. We designed a prototype incorporating this actuation structure, with the main design goals of high functionality, low complexity, robustness, and anthropomorphic sizing. The hand, which features a tendon-driven design, has 15 joints, including 14 dislocatable joints and one revolute hinge joint. The entire hand is powered by a single primary actuator, with two smaller additional motors operating in parallel on the index and thumb. We empirically validated the prototype's performance through qualitative experiments, and its performance was compared to that of other prosthetic devices available on the market through quantitative analysis. Evaluation of the prototype revealed promising results, such as its ability to adaptively grasp various functional objects and execute complex tasks. Force measurements revealed performance comparable to devices on the market. The results indicate that this novel actuation principle, with future refinement, is an interesting new approach to increasing functionality with minimal increase in complexity and can offer an excellent alternative to costly prosthetic devices currently on the market, thereby enhancing the accessibility of functional prosthetic devices for individuals with upper limb loss and improving their quality of life.

This thesis presents the development of a novel Variable Stiffness Gripper using bistable metamaterials (B-VSG) to address limitations in current actuator technologies. The primary objective was to design a lightweight VSG that can switch between different stiffness states without requiring energy to maintain them.

The design process involved optimizing metamaterials for torsional stiffness and bistability. Finite Element Analysis (FEA) and physical experiments were conducted to predict and validate the mechanical behaviour, dynamic behaviour, and stability of the gripper. The results demonstrated a significant increase in torsional stiffness between bistable states, achieving a stiffness increase of 40.2% with a change of 0.173 Nm/rad. The B-VSG showed reduced complexity and lightweight characteristics compared to existing technologies, though further optimization is needed, particularly in the actuation mechanisms and unit cell design.

In conclusion, this research successfully developed a novel, lightweight VSG using bistable metamaterials, proving the feasibility of integrating these materials into robotic actuators. Future work should focus on further improving the metamaterial, enhanced control for stiffness adjustment, and long-term performance tests in real-world conditions.

The challenge of navigating uneven terrain is a critical obstacle in the advancement of robotic
locomotion. Traditional quadrupedal locomotion methods, such as walking, are often
insufficient for dynamic and complex environments. Agile skills like jumping are necessary and
must be adaptable over uneven terrain. This study addresses this issue by developing a policy
for executing jumps over uneven terrain using a single demonstration. Initially, the system
learns to imitate a forward jump based on a single demonstration from a SLIP trajectory
planner. It then generalizes its jumping abilities to various distances in both forward and lateral
directions. The study compares the performance of systems with and without parallel elasticity,
demonstrating the energetic benefits of using elastic actuation for quadrupedal jumping.
Results show that the system with parallel elastic actuation is 15.20% more energy-efficient
and experiences a 15.79% reduction in peak power compared to the system without parallel
elasticity. A policy trained using the proposed methodology successfully performs jumps of
variable distances over uneven terrain with height perturbations of +/-4 cm using only
proprioceptive information.