Independent hardware research from the Pacific Northwest. We build sensing and analysis systems for robotics, ecology, and biosystems — open-source designs, without corporate permission.
Real hardware. Real data. True discovery.
Micro-doppler classifier — not everything that moves is a person
Phase 6 adds a DBSCAN cluster builder and a rule-based micro-doppler classifier to the pipeline. Novel returns are now labeled PERSON, OBJECT, or UNKNOWN before they reach the zone logic. UNKNOWN passes through as PERSON — the classifier can never make the system less safe than the baseline.
Persistent background map — the sensor shouldn't relearn the room every boot
The 15-second background learning phase is necessary — but only once. Phase 5 adds a persistent voxel map saved to disk after the first learn, reloaded on every subsequent boot. Also: a novelty-aware refresh gate that means a stationary person can never be silently absorbed into the background.
Swept-volume workspace clipper — the sensor finally knows what the arm can reach
Phase 7 adds a swept-volume workspace clipper to the pipeline. Detections outside the arm’s reachable envelope are suppressed before they reach zone logic. This is the primary technical differentiator over fixed-mount sensors — and the reason the hardware configuration locked at 3× IWR6843AOP on the forearm link.
Full safety stack live — background learning, ego-motion compensation, and the motionless person problem
The IWR6843AOP is now running a complete ROS 2 safety pipeline on the Jetson — CLEAR/CAUTION/STOP zone detection with ego-motion compensation, voxel background learning, fault handling, and a heartbeat watchdog. One non-obvious bug dominated the session: velocity filtering was silently eating motionless people.
TLV frames live on the Jetson — wrong firmware was the whole problem
Two days chasing a sensor that accepted every config command, then emitted 16 bytes of 0xFF. Root cause: ISK firmware on AOP hardware. The fix was one file.
sensorStart Error -1 — the chirp was running outside the 60–64 GHz band
All 12 calibrations passed. Sensor State 1. Then nothing. The chirp engine silently refused to run because the end frequency exceeded the regulatory band by 750 MHz.
IWR6843AOPEVM arrived — bringup notes and the mmWave Studio trap
EVM in hand, USB detection working immediately. mmWave Studio looked like the obvious next step. It was a trap — the AOP standalone board can’t connect. UniFlash is the right tool.
Project Lumina — something is taking shape
A display project. More when it’s further along.
RadarGuard — mmWave Cobot Safety System Context & Session State Last updated: May 21, 2026 (Phase 7 complete, arm build in progress) Project Identity Product name: RadarGuard Repo name: RadarGuard-mmwave-cobot-safety-system GitHub org: DNTD-Dynamics GitHub URL: github.com/DNTD-Dynamics/RadarGuard-mmwave-cobot-safety-system Repo visibility: PUBLIC as of May 20, 2026 Company: DNTD Dynamics (Snohomish, Washington) Contact: contact@dntddynamics.com Website: dntddynamics.com (GitHub Pages + Cloudflare, repo separate from RadarGuard) License: Business Source License 1.1 (BSL 1.1) Free for non-commercial use (research, education, personal projects) Commercial use requires license from DNTD Dynamics Change date: 4 years from first public release → Apache 2....
RadarGuard — mmWave Cobot Safety System Real-time human presence detection and collision avoidance for robot arms and mobile robots, powered by mmWave radar. Built by DNTD Dynamics · Licensed under BSL 1.1 What is this? RadarGuard is an open-source safety system that uses mmWave radar to detect people in a robot’s workspace and output CLEAR / CAUTION / STOP zone commands in real time. Unlike camera-based safety systems, mmWave radar:...