My name is Ariel Kapusta (Ari). I am a skilled robotics developer, with a passion for investigating and solving the tricky problems faced in robotics.
I currently live and work in Berlin, Germany at Enway GmbH, developing robots for autonomous street sweeping and industrial sweeping.
I earned by Ph.D. in Robotics in 2018 from the Georgia Institute of Technology school of Mechanical Engineering, advised by Dr. Charles C. Kemp in the Healthcare Robotics Laboratory. My Ph.D. dissertation was titled "Task-centric Optimization for Assistive Mobile Manipulators".
In my Ph.D. work I developed exciting new methods and systems, primarily for assisting people with disabilities to perform daily activities (e.g., activities of daily living). I have a passion for addressing both the theoretical and practical sides of robotics problems to see real-world results. My Ph.D. work as well my work at Enway, has covered the entire spectrum of robotics topics. I have covered robotics mechanics, controls, AI, perception, and human-robot interaction extensively. I use the term AI broadly, to include behavioral planning, localization, navigation, machine learning, and motion planning. I have experience designing and implementing from the lowest levels of the robot - hardware, infrastructure, PLC and microcontroller programming, and low-level controllers - up to the most advanced, high levels of robotics programming.
I am excited by the possible real-world benefits from taking robotics from theoretical development into real, practical environments. While it's enjoyable to see robots succeed in laboratory settings, real-world settings and real-world problems are often unique. In my Ph.D. work, I tested my developed methods with real people with disabilities and got to see real impact from my work. In my work at Enway, we are deploying sweeping robots at customers and seeing how the layers of AI work when presented with the messiness of real-world situations. I have seen the disconnect between reality and users in some research robotics, where something like 95% success rates are considered good. Users, customers, real-world scenarios do not work out when the robot behaves poorly (e.g., harms a person) 5% of the time, or only when the environment is particularly dusty. It is exciting to be able to develop from theory to practical application and have to deal with the realities of robotics.
See my CV, Publications, Projects, and Videos pages for more information.
Feel free to contact me via my Contact page.