Skip to content

Latest commit

 

History

4 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SafeSense 3D

Adaptive, Uncertainty-Aware Digital Safety Twin for Automated Machinery

Sense → Estimate → Adapt → Decide → Act

SafeSense 3D is a simulation-first safety architecture for automated machinery that dynamically determines the required protective envelope based on machine motion, hazard proximity, and sensor confidence.

Developed for National Student Hack-A-Stage 2026 — Track 01: Algorithmic Safety & Hazard Mitigation in Automated Systems. Team: Robo Rebels


Problem

Automated machinery commonly relies on predefined safety zones and threshold-based responses. A fixed safety zone does not account for:

  • Machine velocity
  • Stopping distance
  • Reaction delay
  • Sensor uncertainty
  • Occlusion
  • Conflicting sensor observations

The core question: how much safety space does the machine need right now?


Approach

SafeSense answers this continuously by combining machine dynamics with sensor confidence, following a five-stage pipeline:

  1. Sense — virtual LiDAR + thermal observation
  2. Estimate — distance, velocity, and confidence
  3. Adapt — stopping distance + uncertainty margin
  4. Decide — risk classification via TTC
  5. Act — slow / brake / E-stop

Safety principle: uncertainty can expand the protective envelope, but it can never reduce the deterministic safety floor.


Core Model

d_safe = d_stop + d_reaction + d_uncertainty

  • d_stop = v² / (2a) — required stopping distance
  • d_reaction = v × t_delay — sensing, computation, and actuation delay
  • d_uncertainty = f(confidence) — lower confidence increases the protective margin

Risk states are classified using distance, time-to-collision (TTC), and the adaptive safety envelope: NORMAL → WARNING → BRAKE → E-STOP.


Running the Simulation

pip install -r requirements.txt
python3 simulation/run_simulation.py

This runs six deterministic validation scenarios and writes results to results/validation/safesense_validation.csv.


Validation Results

Scenario Confidence Distance (m) TTC (s) d_safe (m) State
Clear / high confidence 0.95 2.75 2.75 1.14 NORMAL
Approach / high confidence 0.92 2.00 2.00 1.17 WARNING
Critical approach 0.90 1.30 1.30 1.18 WARNING
Occlusion / low confidence 0.55 1.65 1.27 1.46 WARNING
Conflicting sensors 0.35 1.25 0.89 1.62 BRAKE
Imminent hazard 0.85 0.82 0.59 1.22 E-STOP

Lower confidence increases the uncertainty margin and therefore expands d_safe. All values above are outputs of the deterministic simulation model, not measurements from physical sensors.


Safety Positioning

SafeSense 3D is a supervisory decision-support layer. It is not a certified industrial safety controller — certified hardwired / safety-PLC functions remain the final safety authority in any real deployment.


Roadmap

  • Calibrate uncertainty models using real sensor data
  • Validate against representative industrial robot/AGV scenarios
  • Integrate with certified safety controllers under appropriate industrial standards
  • Develop automated sensor-placement optimization for complex workspaces

License

See the LICENSE file for licensing information.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages