Steven Haener

- Migrating comms and controls to ROS2 to enable multi-robot collaboration, easy sensor integration, and larger data-stream handling. - Developing an Android-based controller for viewing on-board camera streams and adjusting gait parameters. - Developing building-mapping strategies from on-board cameras and LiDAR using Gaussian Splatting techniques. - Implementing RL policies in IsaacSim for autonomous exploration and traversal of unknown areas.

- Investigating the influence of introducing ground reaction force (GRF) data into the training of bipedal locomotion policies using motion capture data and reinforcement learning pipelines. We hypothesize that including GRF in the locomotion policy will lead to higher sample efficiency and stabler gait for humanoids. - Implementing policies in IsaacSim and MuJoCo for sim-to-sim and sim-to-real transfer on Unitree G1 hardware. - Published a paper on GRF influence in motion retargeting and writing now on objective functions of RL policies.

- Migrated backend communication between robots on the testbed with the server/tracking cameras to ROS2-based client to better handle juggling synchronized tracking, real-time pose streams, and multi-robot coordination. - Led to higher reliability, simpler sensor fusion, and higher bandwidth for large sensor data like cameras. - Upgraded current robot hardware and implemented MATLAB code to include distance sensors (SLAM) and IMUs to enhance the tools provided to researchers that wish to run experiments with a larger arsenal of sensors.

- Held office hours, wrote exams, and graded coursework for a graduate-level AI course covering search, game playing, constraint satisfaction, probability, and Bayes nets. - Performed the same duties for a graduate-level Machine Learning course covering supervised, unsupervised, and reinforcement learning algorithms, pattern recognition through time, and planning under uncertainty.

- Designed and developed a foot in Fusion for a bipedal Cassie robot aimed at improving walking stability, slower walking gaits, and reduced actuation effort, leading to a longer lasting robot and more natural-looking movement. - Integrated tactile/IMU/auditory sensors using C++ with actuated tarsal segments on both sides of the foot. - Developing Python algorithm to leverage these additional signals for classification of the terrain being walked on, informing our control system of the terrain characteristics so it can tune its strategy for increased stability. - Writing a journal paper describing design/testing methodologies and results.

- Served as the engineer that assessed As-Built drawings and proctored Issued For Construction drawings for pipelines, valve automation systems, pumps, compressor stations, and control valves to service tens of thousands of homes and businesses across California, Nevada, Wyoming, and Utah. - Communicated with clients about their needs and tailored drawings to their specific company standards.

- Verified the conditions of building structures and mechanical systems in large commercial spaces to ensure the property buyer understands the scope of repair costs of structures and systems that were not properly maintained. - Prepared technical reports on findings with cost tables and photo logs to provide clients with engineering and regulatory recommendations for the next 12 years to guarantee optimal repair and upkeep of the property.

- Collaborated with a team of mechanical and software engineers towards automating the removal of boulders from farm fields using drone-mounted cameras and an outfitted skid steer, reducing manual labor costs for farmers. - Implemented FEA and design analysis in SolidWorks to reduce the weight of the rock picker by nearly 40%. - Rapidly prototyped using plasma cutting and welding experience and verified successful designs with field tests.

- Designed and developed the thumb module of a robotic rehabilitation device for stroke patients in SolidWorks. - Prototyped and assembled the device using 3D printed components and purchased specialty parts. - Implemented linear actuators, force sensors, and DAQ devices to measure the strength of the patients thumb through different ranges of motion to track recovery progress and possible areas for improvement.

- Designed and developed a machine to measure the strength of wheat stalks with the end goal of crossbreeding the stronger strains of wheat to create a crop that’s more resistant to inclement weather. - Integrated rotary encoders and force sensors to correlate higher force readings per meter with stronger stalks. - Developed data collection and GUI algorithms in C++ and Python to be able to easily use the machine in the field. - Collected field data on various wheat strains to analyze and found stronger breeds to be used in crossbreeding.

Minors in Math, Music, and German | Grand Challenge Scholars Program (GCSP) founding member