Robotic Manipulation

Robots use vision, control software, and robotic arms to pick, place, move, inspect, and operate equipment in production and lab workflows.

Business impact

• Throughput — Ginkgo says its lab runs day and night and uses 100+ RACs, which indicates higher continuous workflow capacity.
• Manual labor required — The cases point to robotic handling of tasks that would otherwise require people, including sample transport, part placement, and maintenance work.
• Task completion speed — Araceli says Endeavor images a 1536-well plate in under four minutes, and Stellantis uses vision guidance to speed robot pick-and-place operations.

Data requirements

• Images from cameras and 3D vision systems (image) — Used to locate objects and read instruments in the robotic workflows.
• Machine-learning data and knowledge databases (tabular) — ALIAS uses machine learning software and access to knowledge gained by other installations.
• Lab imaging data (image) — Ginkgo and Araceli use high-content imaging data for analysis and quality checks.

AI methods and techniques

• Computer vision — Used to locate parts, read cockpit instruments, and analyze images as the robot works.
• Machine learning — Used to learn how to operate aircraft controls and support robot guidance.
• Real-time analysis — Used in the lab workflow to check imaging results as each well is captured and flag issues immediately.
• 3D vision — Used at Stellantis to locate cylinder heads on pallets and guide robot motion.

Industries

Real-world evidence

6 documented case studies on record.

Companies using this: Stellantis, DARPA, Aurora Flight Sciences, BMW, Ginkgo Bioworks, NASA.