SGInnovate × NRP — Tech Talk: “Beam There, Bot”

Notes and reflections from SGInnovate × National Robotics Programme, in collaboration with KABAM Robotics, SIMTech, and ARTC. The event gathered leaders across research and industry to discuss robotics, AI, and talent in Singapore’s deep tech ecosystem.

Overview

Singapore is positioning itself not only as a technology adopter but as a creator of globally competitive robotics solutions. The event emphasized two pillars:

• Technical progress in robotics and AI (perception, control, RL, ROS/ROS2, Sim2Real).
• Talent pipelines and translational mechanisms that turn research into deployable systems.

Keynotes

• Priscilla Looi — Deep learning + robotics direction in Singapore’s ecosystem; building capacity for high‑impact, applied robotics.
• Ai Peng New — Talent as a strategic lever; the need for stronger training, mentorship, and pathways from academia to industry.

Workshops

• Workshop 1 (Dr Lin Wei & Dr Tan Jun Liang): Application‑oriented robotics opportunities; mapping research problems to commercialization and deployment.
• Workshop 2 (Mr Glenn Tan & Ms Sheila Suppiah): Hands‑on development with ROS (and ROS2), emphasizing modularity, scalable integration, and best practices.

Both workshops highlighted bridging foundational research (algorithms, control, perception) with real‑world deployment (interfaces, safety, maintainability).

Panel — Empowering Talent

The highest‑priority segment focused on talent:

• Strengthening pipelines: internships, research residencies, and co‑supervised projects.
• Practical fluency: ROS/ROS2, perception stacks, RL/control, and robust engineering practices.
• Collaboration: academia ↔ institutes ↔ startups for repeatable, reproducible results.
• Open‑source contribution as a credibility signal and accelerant for the ecosystem.

Technical themes I’m tracking

• Multimodal perception: depth + vision + language for intent and feedback.
• RL and planning: growing emphasis on model‑based/world‑model approaches to improve sample efficiency.
• Sim2Real: domain randomization, calibration, evaluation loops, and documentation to make results transferable.
• Systems engineering: data pipelines, safety checks, and benchmarking to move beyond demos.

My next steps

• Build an embodied RL agent with visual servoing in Isaac Gym/MuJoCo; add depth fusion and evaluate transfer.
• Package a ROS2 demo that cleanly integrates perception → control → logging; open‑source the stack.
• Write a short “Singapore Robotics Ecosystem” note (players, labs, datasets, benchmarks) to contextualize projects.
• Seek co‑supervised opportunities linking chemistry + robotics for applied scientific automation.

---

Event link: SGInnovate × NRP — “Tech Talk: Beam There, Bot”