STDA047 September 2026 AWR2188 , AWR2944P , IWR6243 , IWR6843 , IWR6843AOP , IWRL6432
1 Why Radar in Robotics
Cameras are the default perception modality in robotics, but cameras degrade predictably in the scenarios that matter most for safe deployment. Radar is a complementary modality that addresses gaps cameras cannot reliably fill.
| Scenario | Expected Response | |
|---|---|---|
| Camera Alone | Camera and mmWave Radar | |
| Low-light or no-light environment | Degraded or failed detection | Radar performance is lighting-independent, detection maintained where camera degrades |
| Fog, dust, condensate, smoke | Occlusion and missed detections | Radar is significantly more robust to particulates, detection maintained in conditions that challenge camera |
| Glass walls, clear packaging, mirrors | Invisible or misclassified | Glass transmits light rather than reflecting light; polished surfaces deflect light away from the lens. Radar returns from any solid surface regardless of optical transparency |
| Human versus robot versus object classification | Effective in well-lit conditions; degrades at range or under lighting variation | Radar micro-doppler signatures provide an additional classification signal independent of lighting and visual appearance |
| Object velocity measurement | Requires frame differencing, latency and noise increase at low frame rates | Native per-point doppler velocity measurement, no frame differencing required |
| Safety bubble zone enforcement | Limited native velocity data at zone boundary; potential FOV gaps | Range and velocity together support dynamic zone definition and enforcement |
| Factory and warehouse safety | Camera-based systems can degrade under industrial lighting variation, dust, and steam. | Adding a radar layer provides a lighting- and environment-independent detection input with IEC 61508 safety integrity level (SIL) 2 certified hardware integrity |
Figure 1-1 Humanoid Robot Interacting With
Physical Environment