Calculate continuous operational mission duration, battery discharge C-rate, and power trade-off for NVIDIA Jetson AGX Orin 64GB (275 TOPS) running Continuous 4K Object Detection Real-Time (YOLOv10).
With 520 W for the motors and 60 W for the Jetson, the calculator gives 30.1 minutes using 82% of the pack's 355.2 Wh.
With these defaults, running the Jetson at 60 W instead of not at all costs about 3.5 minutes (30.1 versus 33.6 minutes). Propulsion dominates.
It is a gentle load for most LiPo packs. Packs made of cylindrical lithium-ion cells have lower ratings, so check the cell datasheet.
NVIDIA Jetson AGX Orin 64GB is the largest module in NVIDIA's Orin embedded line. It combines a 2,048-core Ampere GPU with 64 Tensor Cores, a 12-core Arm Cortex-A78AE CPU and 64 GB of LPDDR5 memory with 204.8 GB/s of bandwidth. NVIDIA rates it at up to 275 TOPS (INT8, with sparsity) and lets you choose its power budget between 15 W and 60 W through preset modes (15 W, 30 W, 50 W and MAXN). The 60 W default on this page is the top of that range, the right worst case if you run a large detector such as YOLOv10 continuously at high resolution.
On a drone the module needs a carrier board, cameras and cooling, all of which add weight and some power. The developer kit is a desk-sized reference design; flying platforms normally use a smaller third-party carrier board.
The 0.82 factor reserves 18% of the pack. Lithium-polymer cells age quickly when drained flat and their voltage sags under load near the end, so pilots commonly land with 20% or more remaining. The tooltip on the result says 80%; the formula actually uses 82%.
| Quantity | Working | Result |
|---|---|---|
| Pack energy | 16,000 mAh × 22.2 V ÷ 1000 | 355.2 Wh |
| Usable energy | 355.2 × 0.82 | 291.3 Wh |
| Total draw | 520 W motors + 60 W Jetson | 580 W |
| Endurance | 291.3 ÷ 580 × 60 | 30.1 minutes |
| Current and C-rate | 580 ÷ 22.2 = 26.1 A; 26.1 ÷ 16 | 1.6 C |
22.2 V is the nominal voltage of six lithium cells in series (6 × 3.7 V). A 1.6 C continuous load is gentle for most LiPo packs, which are typically rated for many times that. Packs built from cylindrical lithium-ion cells have lower discharge ratings, so check the cell datasheet if you use them.
| Jetson power mode | Total draw | Endurance |
|---|---|---|
| 60 W (MAXN, default) | 580 W | 30.1 min |
| 30 W mode | 550 W | 31.8 min |
| 15 W mode | 535 W | 32.7 min |
| No compute at all | 520 W | 33.6 min |
With these numbers, running the module flat out costs about three and a half minutes compared with no compute at all. Propulsion dominates. The module's weight, and the weight of its cooling, carrier board and cameras, also raises the hover power, and the calculator does not model that unless you increase the motor figure yourself.
Lower power modes reduce the clock speeds and the number of active cores, so the same model runs at a lower frame rate. Whether 30 W is enough depends on your detector size, input resolution and required frames per second. Measure with your own model exported to TensorRT, and note the frame rate in each mode before choosing. Using a smaller YOLO variant or lower input resolution often saves more power than a lower power mode.