Topic 3: Safety, Compliance & Human-Robot Collaboration

Topic 3 introduces the safety and human-collaboration layer required for any robot operating in proximity to people. Rather than treating safety as a bolt-on feature, you will learn to design your system around safety goals, regulatory expectations, and human comfort.

3.1 Industrial Safety Protocols

Safety Functions and Performance Levels

Industrial robots are expected to implement specific safety functions, such as:

• Emergency stop
• Safe torque off
• Safe speed monitoring
• Safe limited position or workspace limits

You will:

• Understand how these functions relate to the idea of performance levels and reliability targets.
• See how safety functions are implemented in hardware (safety relays, dual-channel signals) and software (monitors, interlocks).

Speed Limits and Separation Monitoring

When robots share space with humans:

• Speeds near people must be reduced.
• The robot must stop or slow down when a human enters a defined zone.

You will learn:

• How to conceptually design speed and separation monitoring zones using:
  • Safety scanners or area sensors.
  • Virtual zones defined in your map.
• How to configure robot behavior:
  • Full-speed in “robot-only” zones.
  • Reduced speed in shared aisles.
  • Controlled stop when a human is too close.

Force Thresholds, Safe Posture, and Fall-Recovery

For collaborative manipulation and close interactions:

• Excessive forces or collisions are unacceptable.
• Robots must adopt safe postures when idle or when a fault occurs.

You will:

• Define safe poses (e.g., arms close to torso, low center of mass for humanoids).
• Consider passive and active compliance strategies to limit impact forces.
• Design high-level fall-detection and fall-recovery procedures, including:
  • What to log.
  • When to require human inspection.
  • How to transition back to service.

3.2 Human-Aware Navigation

Personal Space and Social Navigation

Navigation is not just about obstacle avoidance when humans are involved:

• People have expectations about personal space and movement patterns.
• Robots that come too close or move abruptly can feel unsafe.

You will:

• Learn simple models of personal space and how to encode them as costs or constraints in planners.
• Design behaviors such as:
  • Slowing when approaching from behind.
  • Leaving wider gaps when overtaking.
  • Avoiding sudden lateral moves in tight corridors.

Gesture, Gaze, and Multimodal Cues

In some deployments, robots will interpret:

• Hand gestures (e.g., wave to stop, point to a location).
• Gaze or head orientation (e.g., understanding what a person is focused on).
• Voice commands or short phrases.

You will:

• Connect perception capabilities from Chapter 4 with simple human–robot interaction (HRI) behaviors.
• Design clearly visible robot signals (lights, sounds, text displays) so humans can understand what the robot is doing or about to do.

Prioritizing Safety Over Efficiency

Topic 3 emphasizes that throughput is never more important than safety:

• Routes that slightly increase travel time but reduce near-miss incidents are preferred.
• Policy decisions (e.g., always yielding to humans, backing up instead of squeezing through) are intentionally conservative.

You will:

• Explore trade-offs between strict safety policies and system capacity.
• Learn how to document these trade-offs and communicate them to stakeholders.

3.3 Certification & Legal Considerations

Deployment in Workplaces and Public Areas

Deploying robots in workplaces and public spaces involves:

• Local regulations and building codes.
• Employer and facility safety policies.
• Insurance and liability concerns.

You will:

• Learn how to frame a risk assessment for a proposed deployment:
  • Identifying hazards.
  • Estimating likelihood and severity.
  • Defining mitigations and residual risk.
• Understand why signage, training, and physical markings (e.g., painted zones, floor tape) matter for both safety and compliance.

Event Logging, Audit Trails, and Incident Handling

When something goes wrong:

• You must be able to reconstruct what happened.
• Logs and telemetry are essential for diagnosing incidents and demonstrating due diligence.

You will:

• Identify what events must be logged (e.g., E-stop activations, safety zone violations, collisions, near-misses).
• Design black-box style logging that is tamper-resistant and time-synchronized.
• Outline procedures for:
  • Preserving logs after an incident.
  • Conducting internal post-incident reviews.
  • Updating procedures and software to prevent recurrence.