Ravin de Souza

Design and construction of adaptive controllers for dexterous robotic hands based on human principles, with application to common daily tasks. Publishing research results in conferences, journals. Supporting lab activities, conducting demonstrations during lab visits.

I studied grasping with dexterous hands and developed methods to capture principles of human grasp behaviour for transfer to artificial systems. My main contributions were: 1) automatically separating demonstrated grasps into cooperating components, identifying guiding principles for recreating interaction forces. 2) computational framework for task-oriented planning of anthropomorphic hand-arm systems validated using human models and robotic systems for common daily tasks.

Formed and managed team targeting software services for robots. Functioned as technical architect and principal developer while managing team expansion, training and industry/academic relationships. Provided technical consultation on parallel processing of data streams. Robotics software addressed all aspects of manual and autonomous control of a 4-wheeled mobile rover covering environment sensing and modelling from real-time sensors streams, hybrid planning-reactive navigation executive and tele-presence.

Developer support for mobile web applications to Nokia. Created support reference base for mobile browser, documenting design of object-oriented workflow to receive, process, render and interact with web content.

Introduction of new features into the AIX operating system with broad impact such as 1) Optimizing the memory layout of AIX kernel for large (16MB) pages. 2) Adapting the memory management subsystem to exploit distributed physical memory in Non-Uniform Memory Architecture (NUMA) machines. Introduced novel methods for replication and migration of memory transparently to the application (U.S. patent) 3) Autonomous tuning of AIX performance conducted at T.J Watson research center.

Software developer responsible for new feature design and problem resolution in the memory management subsystem of the AIX operating system. Rapidly promoted in view of demonstrated ability to understand fundamental architecture of a complex system (>500000 LOC), to trace root cause of problems from externally visible faults and devise solutions within architectural constraints.

Grasping and Manipulation with dexterous hands, Machine Learning, Estimation and Classification, Optimization.

Computer Architecture, Operating Systems, Algorithms, Computer Networks, Software design