Safety Strategies
Cobots employ several strategies defined in ISO 10218-2:2025 (which incorporates the requirements of the former ISO/TS 15066:2016) to ensure safe human-robot collaboration, specifying protective measures for collaborative industrial robot systems.[60][17] These strategies focus on limiting physical hazards during operation, with updates in the 2025 standard providing enhanced guidance on implementation, including cybersecurity aspects.
One primary strategy is power and force limiting (PFL), which restricts the robot's output to prevent injury upon contact. Under PFL, cobots maintain contact forces and pressures below biomechanical thresholds derived from human tolerance studies, such as maximum quasi-static pressures of 140 N/cm² for the abdomen or 210 N/cm² for the pelvis, with transient contacts allowing up to twice these values (280 N/cm² and 420 N/cm²); for extremities like the hand, quasi-static force limits are 70 N and transient 140 N.[17][61] In practice, whole-body impacts are often limited conservatively to around 80 N to avoid pain or injury. This is achieved through active control systems that monitor and adjust torque in real-time, combined with passive designs like rounded edges and padding.[62]
Another approach is speed and separation monitoring (SSM), which dynamically adjusts the cobot's velocity based on the distance to the human operator to maintain a protective separation. The system calculates a minimum protective distance using factors like human approach speed (assumed up to 1.6 m/s), robot stopping time, and sensor resolution, via the formula S_p = S_h + S_r + S_s + C + Z_d + Z_r (where S_h is human intrusion distance, S_r robot stopping distance, etc.), ensuring the robot halts before collision.[17][63]
Hand guiding enables operators to manually direct the cobot's movements via intuitive handles equipped with force sensors and emergency stops, limiting speed to safe levels (typically under 0.25 m/s) and requiring safety-rated controls to prevent unintended acceleration.[17] Complementing these, safety-rated monitored stop halts the cobot upon human entry into the collaborative workspace, resuming only after clearance is verified, often using presence-sensing devices.[17]
Emergency protocols in cobots include immediate power cutoff upon hazard detection, such as unexpected contact or intrusion, with system recovery times under 1 second to minimize downtime while prioritizing safety.[62] Human-robot interaction zones are defined as collaborative workspaces with dynamic boundaries adjusted in real-time via sensors like cameras or laser scanners, ensuring operations remain within safe perimeters.[17]
Ergonomic considerations enhance safety by reducing musculoskeletal strain during prolonged interactions; for instance, height-adjustable bases allow cobots to align with human postures, promoting neutral body positions and lowering injury risk from repetitive tasks.[64]
Testing protocols for these strategies involve simulated impact tests using biomechanical models to predict injury outcomes, validating that forces and speeds stay within ISO thresholds across various scenarios before deployment.[65]
Risk Assessment Methods
Risk assessment methods for collaborative robots (cobots) follow a systematic approach outlined in ISO 12100, which emphasizes iterative hazard identification, risk estimation, and reduction to ensure safe human-robot interaction before deployment. This standard guides the process by requiring designers to analyze the entire lifecycle of the cobot system, starting with inherent risks from mechanical design and operational tasks. Hazards are categorized through detailed task analysis, identifying potential dangers such as pinching points between cobot links and fixtures, impact forces from unexpected movements, or ergonomic strains from prolonged awkward postures during shared workspaces. For instance, task decomposition involves breaking down operations like assembly or material handling into phases to pinpoint exposure scenarios, ensuring all foreseeable misuse or environmental factors are considered.
Collaborative robot-specific assessments build on these foundations by calculating maximum allowable forces and pressures using biomechanical limits defined in ISO 10218-2:2025 (incorporating ISO/TS 15066), which establishes pain thresholds for various body regions to prevent injury during contact. These limits, derived from empirical studies on human tolerance, specify transient and quasi-static force caps—for example, up to 140 N for hand impacts and 150 N for arm/forearm contacts—to maintain forces below pain-inducing levels without causing harm. Engineers apply these thresholds during the risk estimation phase, often integrating them with power and force limiting strategies to verify that cobot payloads and speeds do not exceed safe interaction parameters under normal and fault conditions. This quantitative evaluation helps prioritize risk reduction measures, such as speed reductions or end-effector redesigns, tailored to the application's variability.[60][62][65]
Simulation tools enhance this process by enabling virtual modeling of human-cobot interactions to predict hazards in diverse scenarios without physical prototyping. Software like Siemens Jack facilitates ergonomic simulations by creating digital human models to assess posture-related strains and collision dynamics, allowing iterative testing of workspace layouts and motion paths. Similarly, the AnyBody Modeling System supports advanced musculoskeletal simulations to evaluate internal tissue loads during potential contacts, providing data on force distribution across body segments. These tools model variables such as operator height variability or unexpected cobot stops, outputting risk metrics like peak force values to inform design adjustments before implementation.[66]
Validation occurs through on-site trials post-simulation, involving real-world measurements to confirm that assessed risks remain below thresholds. Force sensors integrated into the cobot or worn by operators capture impact data during supervised interactions, while operator feedback via surveys or physiological monitoring (e.g., heart rate) gauges subjective comfort and detects unmodeled issues like fatigue. If measurements exceed limits—such as forces surpassing ISO thresholds—designs are iterated, potentially adding padding or recalibrating sensitivity, until residual risk is deemed acceptable. This empirical step ensures the assessment's accuracy across actual operational variability.[67][68]