Research Engineer on Robotics Solutions, AI and image-guided planning algorithms for medical needs

Working towards safe motion planning for high degree-of-freedom robots in unstructured environmentsthrough the use of optimization and learning-based models.- Stochastic Implicit Neural Representations for Motion Planning under Sensing Uncertainty: Sensor uncertainty quantification is posed as a variational inference problem. The uncertainty information is used in a novel chance-constrained hierarchical planner that can be solved efficiently to global optimality using convex optimization. - Optimal Grasps and Placements for Task and Motion Planning in Clutter: Formulation of an optimization-based grounding layer capable of improving the scalability of task and motion planners by jointly optimizing for grasps and object locations in tabletop manipulation problems.- MotionBenchMaker: Open source tool to generate datasets for benchmarking realistic robot manipulation problems.- Human-Guided Motion Planning in Partially Observable Environments: Human preferences are learned by using a Bayesian reward learning approach to learn safe motions. Proposed a method based on sampling-based motion planners and a novel guided trajectory optimization formulation that make learning tractable and efficient.- Robust Motion Planning under Sensing Uncertainty: Formulation of safe motion planning as a trajectory optimization problem where safety is enforced as a positive signed distance between noisy objects and the robot geometry and solved using sequential quadratic optimization and robust optimization.- Teaching assistant for the course Algorithmic Robotics (Fall 2020, Fall 2022)

Taught undergraduate level courses: Robotics (Spring 2019), Analog Electronics (Spring 2018,Spring 2019), Intr. to EE. (Fall 2018), Digital Systems (Spring 2018, Fall 2018), ElectronicsWorkshop (Fall 2018). Advised 3 undergraduate student graduation projects. • Led a 4-month specialized consultancy and assessment project on the analysis of an intelligent debit system using machine learning for Bancolombia. Co-founder of the inter-institutional initiative SinfonIA for working on developing artificial intelligence methods for service robots. Advised undergraduate group SinfonIA for the participation on the RoboCup@Home social standard platform league in Sydney, Australia.

Taught undergraduate level courses: Operating Systems (Spring 2014, Fall 2014, Spring 2015,Fall 2015, Spring 2016, Fall 2016, Spring 2017, Fall 2017), Digital Systems (Fall 2017), Circuits(Fall 2016). Taught graduate level courses: Artificial Intelligence (Fall 2016), Optimization (Spring 2017).Advised 10 undergraduate student graduation projects. Member of the STOx’s team for the development of artificial intelligence methods for mobile robots. Participated in the Small Size League of RoboCup from 2014-2018. In charge of leading the development of robot soccer coordination based on optimization and machine learning. Led the writing of the team’s technical description papers for qualification material in 2014, 2015, 2016, 2017, 2018.

Conducted research of several state of the art topics of the company’s interest, specifically in the field of high performance computing. Involved in the design, development, benchmarking, testing and documentation of the company’s algorithm-based products and prototypes. Provided technical support for a GPU/FPGA/CPU linux-based system. Directly involved in the preparation of several government funding opportunities and patents. Fields of interest to the company include direct and iterative methods for the solution of large scale linear systems, eigenvalue problems and general matrix computations in shared and distributed–memory environments in heterogeneous architectures.

Conducted research of Machine Learning techniques for online pairwise ranking based on graph representations for social network and sports. Designed and implemented novel techniques that attained improved performance over traditional and state-of-the-art techniques in real–life applications. Preparation of one conference-level technical paper.

Marking of home assignments and quizzes, preparation and lecturing of a third year course on optimization and a first year course on fundamental of digital systems.

Design and development of a hardware/software high performance embedded solution for an online learning algorithm into a Virtex II pro FPGA enhanced with a Power PC embedded processor. Full development included hardware and software debugging, documentation and benchmarking. Preparation of a conference-level technical paper submitted and published.