Methodology
How TechMedix predicts failures
The 48-hour prediction window is not a guess. It is the output of a four-stage pipeline that runs continuously across your fleet — from raw telemetry to a scheduled repair.
Telemetry ingestion
TechMedix ingests live robot telemetry — joint temperature, battery state, force/torque, encoder, and vision — from each platform via its native API or edge gateway. Every signal is timestamped and normalized against the platform's documented operating envelope.
Failure-mode signatures
Each platform is modeled against a library of documented failure signatures (e.g. Actuator Overheat: joint temp > 75°C sustained > 30s; Joint Backlash: end-effector tracking error > 15mm). These are the same signatures rendered on every /insights platform page.
Confidence & drift scoring
Signals are scored against baseline behavior. A signature that trends toward its threshold raises a confidence-weighted risk score; we surface source citations and a confidence level (high / medium / low / unverified) so operators know how much to trust each alert.
Lead-time alerting
Because degradation is continuous, TechMedix projects the crossing point and fires an alert with lead time — designed for up to 48 hours before functional failure — so maintenance can be scheduled, not reacted to. Critical signatures short-circuit to immediate dispatch.
Frequently asked
TechMedix is designed to flag impending failures up to 48 hours in advance by modeling continuous telemetry against documented failure-mode signatures and projecting the threshold-crossing point.
Joint/motor temperature, battery state of charge and cell health, force/torque sensor readings, encoder state, and perception/vision feeds where available. Data is ingested via the platform's native API or an edge gateway.
Each platform is mapped to a library of failure signatures with severity, the exact telemetry condition that trips them, and where available a mean-time-between-failures estimate and source citations.
Yes. TechMedix Core is free and open source, published by BlackCat Robotics. The source is available in the BlackCat Robotics repository.
See it on your fleet
Browse the failure modes we track per platform, or book a fleet onboarding call.