Get a PADA Certificate for your sensor or actuator here!
Trusted Perception investigates a simple but consequential idea:
Autonomous systems should be able to perceive hazards and conditions before humans are exposed to them—and people should be able to understand why those perceptions are trustworthy.
Modern autonomous systems can carry increasingly capable sensors and computation into environments that may be dangerous, remote, inaccessible, or difficult for humans to understand directly.
But sensing alone is not enough.
A machine may report that a gas is present, an RF emitter has appeared, an object has moved, a structure has changed, or an environment has become unsafe.
The important question is:
Should anyone trust that conclusion?
Trusted Perception treats perception, cybersecurity, assurance, and human-machine decision making as parts of the same problem.
Mission
Enable autonomous systems to detect, interpret, and explain hazards, conditions, and signals before humans are exposed, with sufficient assurance that humans and other systems can trust the resulting perception.
Trusted Perception follows several principles:
Start with the decision, not the sensor.
Rather than asking, “What can we build with this sensor?” begin with:
What important decision is currently being made with inadequate perception?
Use existing platforms whenever practical.
Robots, drones, vehicles, fixed installations, and distributed systems can serve as carriers for perception technology. Building another robotic platform is not necessarily the problem that needs to be solved.
Experiment before committing.
Small, inexpensive experiments should reduce uncertainty before substantial technical or financial commitments are made.
Security and assurance are part of perception.
A perception system that can easily be spoofed, manipulated, misunderstood, or used outside its validated conditions cannot be considered trustworthy.
Build knowledge before products.
Trusted Perception is intended to develop over multiple years. Research, experimentation, intellectual property, partnerships, teaching, licensing, consulting, or commercial products may eventually emerge—but only when supported by evidence.