Architecting Multithreaded Robotics Applications in C++
summary
Claim: a C++ tool named Lambda C++ library hides threads so robotics engineers think linearly while using multicores. It implements pre‑parallelized linear algebra, Python bindings, a benchmark/visualizer, and ROS2 integration. Evaluation on a Core i7 6‑core machine shows performance comparable to OpenMP, best results for a 2000×2000 multiply using four threads with affinity and priority, and lower deviation in worst‑case times; expert interviews (six) report willingness to adopt while asking for more ops and real‑time/synchronization.
AI-written summary · about
chapters (19)
- 0:00Multithreaded robotics in C++
- 3:52Software architecture role
- 7:01C++ robotics frameworks
- 8:06Robotics app problem
- 12:19Multithreading challenges
- 14:50Problem reformulation
- 17:14Existing solutions shortfalls
- 19:04Lambda framework concept
- 22:10Lambda parallelization API
- 23:59Benchmarking Lambda
- 26:28Hardware awareness & visualization
- 27:33Lambda math operations
- 28:37Lambda with ROS2
- 29:40Evaluation setup & metrics
- 30:24Quantitative results
- 32:32Load deviation analysis
- 34:00Expert feedback
- 35:02Lambda future summary
- 35:44Q&A
related talks
01h 22mCostless Software Abstractions for Parallel Architectures
59mC++ on its way to exascale and beyond
01h 16mPaying for Lunch: C++ in the ManyCore Age
01h 06mAsynchronous Computation in C++
01h 02mOverview of Parallel Programming in C++
48mConsiderations for choosing the parallel computing strategy
30mBridging robotics and systems programming:Why Copper is a game changer
01h 03m...Scaling Visualization in concurrent C++ programs