Battery-aware MPC adaptive cruise control: live demo
Give it a lead-vehicle speed trace. It runs the paper's frozen BMW i3 energy model with a conventional ACC and the battery-aware MPC, and reports Wh/km, gap and safety.
1. Lead-vehicle speed trace
Paste one number per line (speed), or two columns time,speed (CSV). 30 to 180 s of driving.
2. Settings
Results
Speed
Gap to lead
What this is, and what it is not
This runs the same simulation code as the paper's open package, on your trace. It is a research demo, not a vehicle controller and not a range estimate for any real trip.
The example trace is synthetic and deliberately extreme stop-and-go, so its saving is much larger than the paper's result on real traffic (median -2.83% for the oracle ceiling, 160 real windows). Do not quote it as a finding.
The energy model is one frozen BMW i3 (1270 kg, regen efficiency and cap as set above); other cars will differ.
"Oracle" preview is an upper bound on what MPC can gain; it assumes the lead's future speed is known. "Naive" is deployable but gains less.
Speed breakers use the same stylised scenario as experiment S1b: the lead slows to about 15 km/h at each breaker and the map-aware ACC and MPC see their positions. It is a stylised test, not measured Indian traffic, and your trace is not the paper's data, so the number is not a paper result.
Limits: 30 to 180 s per run, lead speed 0 to 46 m/s, 20 breakers; the ego car is capped at 38 m/s. Runs take roughly 10 to 25 s (longer traces, and the first run after a quiet spell, are slower). The service is rate-limited and may refuse bursts.
Uploaded traces are processed to produce the result and are not stored; request logs are kept for 14 days.