Robot trajectory planning with hybrid quantum algorithms
summary
AWS and BMW build an end-to-end robot trajectory pipeline for sealant robots that compares a QUBO→QBSolv→D‑Wave annealing path against a native random‑key approach: a BRKGA implementation and a novel RKO‑DA (random‑key optimizer with dual‑annealing) decoder. Benchmarks use a 52‑seam “large” (with occlusions) and 71‑seam XL dataset; RKO‑DA cuts the greedy baseline from 37.06s to 33.30s (~3.75s, ≈10%) on large and by 2.39s (≈3.5%) on XL, and BMW adopts the code for production.
AI-written summary · about
chapters (19)
- 0:00Introduction and teams
- 3:19Business use case explained
- 5:54Why the problem is difficult
- 7:18Mathematical formulations overview
- 8:02QUBO formulation details
- 10:04Applying QUBO to robot motion
- 12:53Handling constraints in QUBO
- 14:41Native formulation advantages
- 15:44BRKGA solver introduction
- 17:33Simulated annealing explained
- 20:24Quantum annealing principles
- 22:57QBSolv for quantum hardware
- 23:41BRKGA components and process
- 27:18Handling multidimensional nodes
- 30:12Random-key optimizer with dual annealing
- 32:00Benchmark datasets and results
- 34:49Scalability and runtime analysis
- 39:20Statistical solver comparison
- 41:33Summary and future outlook
related talks
14mQuantum Computing- Exploring Automotive Applications
17mExploring Quantum Algorithms for Optimal Sensor Placement in Production Environments
19mExploring Quantum Advantage for the German industry
22mQUARK: A Framework for Quantum Computing Application Benchmarking
21mAutomated Scheduling in the Healthcare Sector
41mQuantum Computing in Practice: Where We Are, What Works, and What’s Next
28mCloning And Beyond: A Quantum Solution to Duplicate Code
20mQuantum Computing as a Service- Cloud-based and automized QML Solu