Validation environments
Test systems that leave a usable record.
Herder builds the target setup, deployment path, data capture, and analysis workflow needed to run a technical test again and explain the result.
The problem
A plot is not evidence if no one knows the software revision, target state, input conditions, or instrument setup that produced it. A bench process that works only for one operator is not a reusable test system. The work is to make the path from deployment to result clear and repeatable.
The work
| Target operation | Linux setup, SSH, process control, filesystem work, device access, and target recovery. |
|---|---|
| Deployment | Artifact cleanup, host-to-target transfer, model lifecycle control, status checks, and result retrieval. |
| Test data | Instrumentation, CAN and serial logs, encoder capture, structured test inputs, and timestamped records. |
| Analysis | Python, MATLAB, and Simulink tools for plots, scaling checks, operating-point work, and stated test limits. |
Evidence
A host-to-target workflow for physical motor-control experiments. It cleaned deployed artifacts, inspected the target, started and checked control models, retrieved logged data, and supported repeatable work over SSH. The target used a TI AM3358 Sitara ARM Cortex-A8 running Debian with MATLAB/Simulink, PWM motor drive, TB6612FNG drivers, and quadrature encoder feedback.
“Fast and thorough setup of Linux systems; diagnosed and solved a manufacturer issue with parts. Great service overall!”