Singapore
I have close to a decade of experience in the field of robotics, from research to industry. I am from a mechanical engineering background and transitioned to robotics early on. Key expertise topics include 2D and 3D robotics vision, for robotic arms and mobile robot. I have executed POCs and Industrial System development, from sales to technical execution and delivery.
- Eureka 3D camera Product Owner (Software, Hardware, R&D) - Europe business development
- Engineering role Technical contributions, team management, project management of Proof of Concepts and Industrial Projects. Key Industries: Aerospace, Manufacturing, Logistics Key areas: 2D/3D vision, calibration, industrial robotic arms, mobile robots - Customer-oriented role Inquiry assessment, technical scoping, pricing, sale negotiation, product demonstration and training, in South East Asia and Europe. Total deals generated > 0.7M SGD.
R&D on 2 kind of robotic applications : - 3D mapping & localisation for autonomous mobile robots (indoor, outdoor) - Dexterous manipulation & control (robotic arms) Awards: KUKA Innovation Award 2021 finalist Best Demo Award @AAAI 202 Best Live Presentation Award @ICSR 2021 Best Poster Award @A*STAR Scientific Conference 2021
R&D on 2 kind of robotic applications : - 3D mapping & localisation for autonomous mobile robots (indoor, outdoor) - Dexterous manipulation & control (robotic arms)
Develop computer vision solutions for the 5-year Collab-AI project (A*STAR I2R, A*STAR IHPC, A*STAR ARTC, NUS). Objective : collaborate and teach new tasks/objects to a cobot as one would do with a human operator. * Main architect (visual perception side) of the prototype we presented at the KUKA Innovation Award 2021: Artificial Intelligence Challenge (Team CHRIS). Developed a fully working and integrated solution within 4 months. Selected has one of the 5 best concepts. - software architecture (ROS), code refactoring (C++ 11 & above, Python), enforce good software engineering practices (Git, Docker) - RGB-D camera calibration (intrinsics & extrinsics, for eye-in-hand and eye-to-hand) - object detection (Faster R-CNN, Pytorch) & 3D object position estimation - hardware design and setup - project management (~7 team members) - main demo presenter (to the jury and public) * R&D for the longer term project (5yrs) - refactoring of raw research code to integrate into the robot and enable cross-platform compatibility (modern C++, Python, ROS1) - integrate object and gesture recognition modules (Faster R-CNN, Pytorch) - integrate visual grounding module, task teaching from visual demonstration (modern C++) - integrate speech transcription solution (Python) - create a GUI for visual feedback (Tkinter) - design software architecture & enforce best practises (compatible libraries, OS, version control, cross-platform compatibility) - benchmark cloud computing setup (AWS) for ML training and inference - promote cross-team collaboration __________________ A*STAR is the Singapore R&D Agency. The institute under which I am working, I²R, performs research in areas including Visual Intelligence and Robotics & Autonomous Systems. As a research engineer, I contribute to research efforts and transform raw ideas into working prototypes.
IPAL is a French-Singaporean joint research laboratory operating as one of CNRS’ (French National Center for Scientific Research) overseas unit. They are working with local research actors such as A*STAR or NUS. The challenge I addressed during this internship (which is somehow a master's thesis) was to establish a mapping between fixed grasping policies and geometries represented by coefficients coming from their mathematical representation : B-Splines. The idea was to be able to identify an object on a table, thanks to a 3D camera, in order to grasp it with a robotic arm later. I started working on 3D point clouds coming from the camera. I had to isolate the desired object (which was a bottle), and then assemble different point clouds coming from different camera angles to recreate the full object. This highly helped me to get familiar with C++ libraries such as PCL. Later on, I focused on surface reconstruction from the point clouds. I didn't use a triangular mesh because the aim was to work with compact and precise surface representation. Hence, I worked on B-Splines and NURBS, through C++ librairies using these tools. The final goal was to establish a mapping between a local geometry on the perceived object and a grasping policy, to grab the object smoothly. Due to lack of time, this part hasn't been implemented. Bonus : once again, I was in charge of organizing football games against other startups working nearby.
Wandercraft, is a French startup that develops, builds and commercialises a lower-limbs exoskeleton to help people with disabilities to walk again. In this project, within the mechanical conception team, I created and optimized different parts of the new robot. This included the redesign of the back assembly, the foot assembly, and plastic shells. I also carried out FEM simulations to comply with safety requirements. Thanks to this modifications, a new prototype of the exoskeleton has been successfully tested on patients. Bonus : during my internship, I was appointed captain of the football team - I had to plan games against other Parisian startups.
Tabobine is a young Parisian startup that develops, owns and installs vintage photo cabins throughout festive areas in Paris. When I arrived, the company had 30 cabins running. The aim was that by the end of my internship, Tabobine could launch a new production of 30 cabins, that would be more autonomous, stronger, and cheaper, while looking the same. In this project, I have been in charge of the conception and design of this new product. It mainly involved CAD-drawing, prototyping, project management and manufacturing