Wearable on-body robotics for haptics or for muscular assistance require compliant, efficient, fast, lightweight and high force actuators. I will present flexible actuators developed in my lab that operate on electrostatic principles, which can offer high speed and high energy density, but (for now) lower force than pneumatics.  The electrostatic principle provides a clear path to untethered operation. I will report both fiber-format as well as patch-format devices, the former aimed at soft exoskeletons, the later for cutaneous haptics.

We all love to report impressive numbers for metrics such as efficiency and specific power. I will give a few examples of how our, often unstated, assumptions lead us to claim performance metrics different by orders of magnitude from reality. I close by suggesting a few routes to standardize our reporting.

Advancing robot manipulation implies that you are able to measure the state-of-the-art performance and then outperform it. To do so, objective performance metrics and evaluation methods are needed to benchmark robot manipulation capabilities. While many advancements have been made to develop common benchmarking tools, testbeds, protocols, and metrics, the high dimensionality of a robot manipulation system and the non-deterministic nature of their algorithms make true side-by-side comparison very challenging. Comparing results from one lab to those collected in another lab can be very misleading due to variances in hardware and environmental conditions. Instead, you can attempt to reproduce results from the other lab in your own lab to generate your own baseline for comparison, but we do not yet have the level of modularity and interoperability needed in order to easily facilitate reproduction. This talk will review the current state of reproducibility for “rigid” robot manipulation (as a counter to “soft” robotics), and how efforts like the COMPARE Ecosystem project aims to improve it. 

Creating life-like robots that can perform dexterous tasks in the real world remains a central challenge in robotics. At the Soft Robotics Lab, our broader mission is to develop robust musculoskeletal and biohybrid robotic systems capable of adaptive, real-world interaction. In this talk, we will present ORCA, an Open-source, Reliable, Cost-effective, Anthropomorphic robotic hand designed to directly address the reproducibility challenges that often hinder progress in soft and compliant robotics. ORCA is a tendon-driven, 3D-printed platform featuring 17 degrees of freedom, reliable auto-calibration capabilities, and a fully open-source software and hardware stack. From mechanical design to electronics and control, every aspect of ORCA has been developed with reproducibility in mind, enabling consistent replication across multiple labs and supporting high-throughput experiments that range from robust teleoperation and accurate control to repeatable imitation and reinforcement learning policy rollouts. We will discuss how ORCA contributes to a more transparent and scalable research ecosystem, including shared CAD files, simulation models, assembly guides, and benchmarking experiments. This work exemplifies how open, community-oriented design can help overcome the reproducibility bottlenecks in soft robotics and facilitate broader adoption of dexterous robotic platforms.