Soft robots and particularly inflatable robots are of interest because they are lightweight, compact, robust to impact, and can interact with humans and their environment relatively safely compared to rigid and heavy traditional robots. Improved safety is due to their low mass that results in low-energy collisions and their compliant, soft construction. Inflatable robots (which are a type of soft robot) are also robust to impact and have a high torque to weight ratio. As a result inflatable robots may be used for many applications such as space exploration, search and rescue, and human-robot interaction. One of the potential problems with inflatable or pneumatically actuated robots is air leaking from the structural or actuation chambers. In this thesis methods are demonstrated to detect leaks in the structural and actuation chambers of inflatable and pneumatically actuated robots. It is then demonstrated that leaks can be slowed by lowering a target pressure which affects joint stiffness to prolong the life of the system. To demonstrate the effects of lowering the target pressure it is first shown that there exists a trade-off between the commanded target pressures at steady-state and the steady-state error at the robot end effector under normal operation. It is then shown that lowering the target pressure (which is related to stiffness) can extend the operational life of the system when compressed air is a limited resource. For actuator leaks a lower target pressure for the leaking joint is used to demonstrate the trade-off between slowing the leak rate and system performance. For structural leaks a novel control algorithm is demonstrated to lower target pressure as much as possible to slow the leak while maintaining a user specified level of accuracy. The method developed for structural leaks extends the operational life of the robot. Long-term error during operation is decreased by as much as 50% of the steady-state error at the end effector when compared to performance during a leak without the control algorithm. For actuation leaks in a joint with a high-torque load the possibility of a 30% increase in operation time while only increasing steady-state error by 2 cm on average is demonstrated. For a joint with a low-torque load it is shown that up to a 300% increase in operation time with less than 1 cm increased steady-state error is possible. The work presented in this thesis demonstrates that varying stiffness may be used to extend the operational life of a robot when a leak has occurred. The work discussed here could be used to extend the available operation time of pneumatic robots. The methods and principles presented here could also be adapted for use on other types of robots to preserve limited system resources (e.g., electrical power) and extend their operation time.

<p>The soaring demands for always-on and fast-response online services have driven modern datacenter networks to undergo tremendous growth. These networks often rely on scale-out designs with large numbers of commodity switches to reach immense capacity while keeping capital expenses under check. Today, datacenter network operators spend tremendous time and efforts on two key challenges: 1) how to efficiently utilize the bandwidth connecting host pairs and 2) how to promptly handle network failures with minimal disruptions to the hosted services.</p><p>To resolve the first challenge, we propose solutions in both network layer and transport layer. In the network layer solution, We advocate to design practical datacenter architectures for easy operation, i.e., an architecture should be reliable, capable of improving bisection bandwidth, scalable and debugging-friendly. By strictly following these four guidelines, We propose DARD, a Distributed Adaptive Routing architecture for Datacenter networks. DARD allows each end host to reallocate traffic from overloaded paths to underloaded paths without central coordination. We use congestion game theory to show that DARD converges to a Nash equilibrium in finite steps and its gap to the optimal flow allocation is bounded in the order of 1/logL, with L being the number of links. We use a testbed implementation and simulations to show that DARD can achieve a close-to-optimal flow allocation with small control overhead in practice.</p><p>In the transport layer solution, We propose Explicit Multipath Congestion Control Protocol (MPXCP), which achieves four desirable properties: fast convergence, efficiency, being fair to flows with different RTTs and negligible queue size. Intensive ns-2 simulation shows that MPXCP can quickly converge to efficiency and fairness without building up queues despite different delay-bandwidth products.</p><p>To resolve the second challenge, recent research efforts have focused on automatic failure localization. Yet, resolving failures still requires significant human interventions, resulting in prolonged failure recovery time. Unlike previous work, we propose NetPilot, a system aims to quickly mitigate rather than resolve failures. NetPilot mitigates failures in much the same way operators do -- by deactivating or restarting suspected offending components. NetPilot circumvents the need for knowing the exact root cause of a failure by taking an intelligent trial-and-error approach. The core of NetPilot is comprised of an Impact Estimator that helps guard against overly disruptive mitigation actions and a failure-specific mitigation planner that minimizes the number of trials. We demonstrate that NetPilot can effectively mitigate several types of critical failures commonly encountered in production datacenter networks.</p> / Dissertation