Robot Collision Detection: Standards, Testing Methods, and Safety Compliance

Robot collision detection is a critical safety technology that enables robotic systems to identify physical contact with humans, obstacles, or equipment and respond instantly to prevent injury and damage. As collaborative robots and mobile robots increasingly share workspaces with people, compliance with international safety standards such as ISO 10218 and ISO/TS 15066 has become essential for market access. This article presents a comprehensive overview of robot collision detection principles, applicable standards, testing samples, testing items, test methods, and typical industrial applications.

Technical Overview

Robot collision detection is the capability of a robotic system to identify unintended physical contact between the robot and a person, obstacle, workpiece, or another machine, and to trigger an appropriate safety response such as an immediate stop, a controlled retreat, or the limitation of contact forces to biomechanically safe levels. As robots increasingly operate in shared workspaces with human operators—particularly in collaborative and mobile applications—reliable collision detection has become a core requirement of machinery safety compliance and a decisive factor in risk assessment.

From an engineering perspective, collision detection spans both pre-contact avoidance and post-contact reaction. Pre-contact strategies rely on external perception devices to maintain a protective separation distance, while post-contact strategies rely on internal measurements such as joint currents, joint torques, and wrist forces to detect and mitigate an actual impact. Both strategies must be validated through structured testing, often performed by accredited third-party laboratories or manufacturer test facilities, to demonstrate conformity with the relevant international standards.

Detection Technologies

Modern robot collision detection is implemented through several complementary techniques:

• Model-based torque estimation: disturbance observers compare commanded motor currents with a dynamic robot model and flag a collision when the residual exceeds a defined threshold.
• Joint torque sensors: direct measurement of external torque at each joint provides high sensitivity without additional external hardware.
• Six-axis force/torque sensors: wrist- or base-mounted sensors measure contact forces applied to the end-effector or tool.
• Capacitive and pressure-sensitive robot skins: distributed sensor layers covering the arm surface enable contact localization and pre-contact proximity sensing.
• External perception systems: safety-rated laser scanners, LiDAR, and 3D vision systems support speed and separation monitoring (SSM) before contact occurs.
• Geometric collision checking: software algorithms based on bounding volume hierarchies and swept-volume computations verify collision-free trajectories during path planning.

Standards Framework

• ISO 10218-1 and ISO 10218-2 define safety requirements for industrial robots and robot systems.
• ISO/TS 15066 specifies collaborative robot operation, including power and force limiting (PFL) and biomechanical limits for human body regions.
• ISO 13849-1 and IEC 62061 govern the safety-related parts of control systems, defining Performance Levels (PL) and Safety Integrity Levels (SIL).
• IEC 61496 covers electro-sensitive protective equipment used for contactless collision prevention.
• ISO 13855 establishes the positioning of safeguards based on approach speeds and stopping performance.
• ISO 3691-4 addresses safety requirements for driverless industrial trucks, including mobile robots.
• GB 11291.1 and GB/T 36008 provide corresponding Chinese national standards for industrial robots and collaborative robots.
• ASTM F45 series standards address performance measurement methods for robotic and autonomous systems.

Testing Samples

Collision detection testing is performed on complete robotic systems as well as on individual safety components. The most common sample categories are described below.

Collaborative Robot Arms

Collaborative robots (cobots) with typical payloads ranging from 3 kg to 30 kg are tested for power and force limiting performance, including contact force, contact pressure, and detection sensitivity across the entire arm surface and operating envelope.

Industrial Manipulators

Six-axis articulated arms, SCARA robots, and delta robots equipped with external safety devices are evaluated for stopping performance, safety-rated soft axis and space limiting, and the correct interaction between the manipulator and its protective equipment.

Mobile Robots

Automated guided vehicles (AGVs) and autonomous mobile robots (AMRs) fitted with safety laser scanners, mechanical bumpers, and contact strips are tested for obstacle detection, protective field switching, and collision response behavior in mixed human-vehicle traffic.

Humanoid and Service Robots

This emerging category includes humanoid robots, quadrupeds, and delivery robots that operate in public or semi-public spaces, where whole-body collision detection and gentle force limiting are essential for safe human interaction.

Safety Components and Peripherals

Collision sensors, tool changers, grippers, force-limiting end effectors, and controller-integrated safety functions such as safely-limited speed and safe stop are individually characterized to confirm their contribution to the overall safety function.

Testing Items

Typical test items for a comprehensive collision detection evaluation include:

• Quasi-static contact force: sustained clamping or pressing forces measured against the permissible limits defined for each body region.
• Transient contact force: peak impact forces recorded during dynamic collisions at defined speeds and payload conditions.
• Contact pressure distribution: localized pressure on skin-equivalent surfaces to verify compliance with pressure thresholds.
• Collision detection response time: elapsed time between contact onset and the triggering of the safety stop.
• Stopping time and stopping distance: measured from the detection signal to a complete standstill of the robot.
• Protective separation distance: validation of speed and separation monitoring parameters for contactless systems.
• Detection coverage: confirmation that the entire robot body, not only the end-effector, is monitored.
• Threshold sensitivity: balancing reliable detection against false alarms caused by normal process-related forces.
• Safety function integrity: verification of PL d or SIL 2 ratings for the collision detection safety function.
• Repeatability and robustness: consistent performance across repeated impacts, orientations, speeds, and payloads.

Testing Methods

Collision detection testing combines physical measurement, functional safety assessment, and analytical verification. The principal methods applied by accredited third-party laboratories and manufacturer test facilities are outlined below.

Force and Pressure Measurement

Calibrated load cells and pressure-indicating films or mats are applied to test devices whose stiffness and damping characteristics reproduce the compliance of specific human body regions, as defined in ISO/TS 15066. The robot executes quasi-static clamping and transient impact sequences, and the resulting forces and pressures are compared against the applicable biomechanical limits.

Impact Testing

Instrumented impact rigs use a defined combination of effective mass, spring, and damper parameters representing each body region. The robot approaches the test device at its maximum declared speed, and the transmitted peak force, contact duration, and energy transfer are recorded. Testing is repeated at multiple speeds, tool orientations, and payload conditions to establish the safe operating envelope.

Stopping Performance and Response Time Measurement

Stopping time and stopping distance are measured from the moment the collision detection function triggers to the complete cessation of motion, in accordance with the measurement provisions of ISO 10218-1. These values feed directly into separation distance calculations per ISO 13855 and into the overall safety function validation.

Speed and Separation Monitoring Validation

For systems relying on contactless collision prevention, target approach tests verify the correct functioning of safety-rated scanners or vision systems, including protective field switching, response times, and output signal behavior in line with IEC 61496.

Functional Safety Assessment

The collision detection safety function is evaluated according to ISO 13849-1 and IEC 62061, including fault injection testing, analysis of the safety control architecture, and verification of the achieved Performance Level or Safety Integrity Level. Simulation and swept-volume analysis may supplement physical testing to confirm full coverage of the robot working envelope.

Applications

• Automotive manufacturing: collaborative machine tending, screwdriving, and assembly where operators and robots share workstation boundaries.
• Electronics and precision assembly: force-limited collaborative robots handling delicate components in close proximity to human workers.
• Logistics and warehousing: mobile robots operating in mixed human-vehicle traffic and relying on scanner-based collision prevention.
• Machining and surface finishing: sanding, polishing, and welding applications with continuous contact force regulation.
• Healthcare and laboratory automation: robotic assistants and laboratory platforms subject to strict force and pressure constraints.
• Service and hospitality: delivery robots and humanoid robots navigating public spaces with certified collision safety.
• Regulatory compliance: test evidence supporting CE marking under the EU Machinery Regulation and conformity with GB standards for the Chinese market.
• Research and development: benchmarking prototype detection algorithms and sensor configurations prior to market release.

FAQ

What is the difference between collision avoidance and collision detection?

Collision avoidance is proactive: it prevents contact using perception systems such as laser scanners and vision before a collision can occur. Collision detection is reactive: it identifies an actual physical contact through torque, current, or force signals and triggers an immediate safety response. Most compliant robot systems combine both strategies.

Which standard defines the force and pressure limits for human-robot contact?

ISO/TS 15066, the technical specification for collaborative robot safety, defines biomechanical limits for quasi-static and transient contact forces and pressures for each body region. These limits serve as the primary acceptance criteria in collaborative robot collision testing.

Is collision detection testing mandatory?

Robots placed on the market as machinery must undergo conformity assessment, including risk assessment and validation of safety functions. For collaborative applications claiming power and force limiting, documented collision testing is effectively mandatory to substantiate safety claims. Many manufacturers and integrators also engage accredited third-party laboratories for independent verification.

How do quasi-static and transient contact differ in testing?

Quasi-static contact refers to slow, sustained clamping or pressing, such as a body part trapped between the robot and a fixture, and is evaluated against force and pressure limits over time. Transient contact is a brief dynamic impact, evaluated by peak force and energy transfer. Both scenarios must be tested because their injury mechanisms differ.

Can collision detection replace physical guards and fences?

Under specific collaborative operating modes and a documented risk assessment, validated collision detection and force limiting can justify the removal of physical barriers. However, this is permissible only when all foreseeable contact scenarios remain within the biomechanical limits of the applicable standards and when speeds, payloads, and separation distances are adequately controlled.

How often should collision detection be revalidated?

Revalidation is required after design changes, safety-relevant software updates, changes in payload or tooling, or modifications to the workcell layout, and should also be performed periodically as part of a maintenance regime to confirm that detection thresholds and stopping performance remain within specification.