Robots are increasingly being used to complete tasks in complex environments. To complete these tasks, two of the most fundamental choices that the robot must make are (1) task planning: which tasks to complete and in what order, and (2) path planning: what path to take when navigating between those tasks. In this seminar, I […]
Despite the recent progress in robot learning, robotics research and benchmarks today are typically confined to simple short-horizon tasks. However, tasks in our daily lives are much more complicated — consisting of multiple sub-tasks and requiring high dexterity skills — and the typical “learning from scratch” scheme is hardly scale to such complex long-horizon tasks. […]
The promise of multi-robot systems is that they can complete tasks quicker and more efficiently than a single robot. However, such performance gains are only realized if robots can communicate with each other and coordinate their actions. As on-board autonomy advances, robots are increasingly being deployed in environments without existing wireless infrastructure and must rely […]
Modular robots offer the promise of flexibility; we will be able to deploy teams of generalist modules and reconfigure them into various specialists to perform whatever task is necessary. Surface swimming aquatic modular robots in particular are not well studied, however, because the literature assumes that the modules need to be holonomic, which makes them […]
This dissertation discusses my work in building an HRI platform called Quori and my once separate now integrated work on a manipulation method that can enable robots like Quori, or any more capable robot, to move large circular cylindrical objects. Quori is a novel, affordable, socially interactive humanoid robot platform for facilitating noncontact human-robot interaction […]
Variable Topology Trusses (VTTs) are a new type of modular, self-reconfigurable robot (MSRR). Like other self-reconfigurable robot systems, a VTT can change its shape to adapt to a variety of tasks, offering a wider range of capabilities than a single robot configuration. While most MSRRs consist of robot modules that are connected in a chain […]
With the advent of more capable and cheaper legged robots, robotic legged locomotion is becoming an increasingly attractive solution for locomotion and embodiment. Legs offer expanded capability over wheeled robots, albeit at the cost of complexity and efficiency. For example, walking robots explore rubble-filled caves, inspect industrial sites, move around human spaces (e.g., stairs), and […]
Materials with electroprogrammable stiffness and adhesion can enhance the performance of robotic systems but achieving large changes in stiffness and adhesive forces in real time is an ongoing challenge. Electroadhesive clutches can rapidly adhere high stiffness elements, although their low force capacities, high activation voltages, and inability to separate and turn off stiffness changes reliably […]
Different from neuron-based computational intelligence through the brain, physical intelligence leverages structural designs and smart materials to physically encode sensing, actuation, control, adaption, and decision-making into the body of an agent. The stimuli-responsive body materials can enable autonomous sensory, actuation, powering, and other physical intelligence functions. The structural designs of soft body can simplify the […]
The CIS Department and GRASP Lab invite you to please join us on Thursday, November 17th, at 3:30pm as we celebrate the life and legacy of Dr. Max Mintz, Professor of Computer and Information Science. Max joined Penn as an assistant professor of Systems Engineering (now part of ESE) in 1974. He changed his primary […]
