Multi-Vehicle Map Fusion Using GNU Radio
summary
An open-source ERA application fuses 2D occupancy grids across vehicles using a GNU Radio implementation of DSRC, performs V2V map exchange and fusion, and profiles the DSRC transceiver on a Xilinx UltraScale+ ZCU102 SoC. Two hotspots—complex exponentiation and the Viterbi butterfly—receive optimizations that measure a 24% throughput improvement with projected gains up to 58%; FPGA and NEON implementations cut cycles per complex-exp from ~37 to ~5–7 and eliminate CPU bottlenecks.
AI-written summary · about
chapters (15)
- 0:00Multi-vehicle map fusion intro
- 1:00Swarm intelligence paradigm
- 3:48Why swarm intelligence now
- 4:53Car-centric AV limitations
- 6:36DARPA EPOCHS project
- 8:42ERA app details
- 10:4811p transceiver in GNU Radio
- 11:50Transceiver bottlenecks
- 13:57Hardware acceleration of exponent
- 14:59ARM NEON vectorization
- 15:39Future accelerator optimizations
- 17:25Over-the-air validation
- 18:47Summary
- 20:10ERA as benchmark platform
- 21:21Q&A
related talks
36mGNU Radio at the Allen Telescope Array
27mAzure Orbital, RF Simulation, and Spectrum Processing on Azure
33mOpen Source V2X Library
13mUtilizing Vehicle Trajectory Data to Characterize Performance Measures of a Highway Corridors
19mSensor-based Flow Optimization on connected real-world intersections via a SUMO Feature Gap Analysis
20mSimulating platooned connected autonomous vehicle in a large scale urban scenario
16mExploring the Potential of Zephyr in Automotive and Software Defined Vehicles
17mSignal priority for improving fluidity and decreasing fuel consumption