External Torque Sensors vs Joint Current Sensing: Collision Detection

External torque sensors excel at sensitivity and accuracy because they directly measure contact forces at the mechanical interface, independent of the motor's electromagnetic noise. For example, a 6-axis force-torque (FT) sensor mounted at the robot wrist can detect contact forces as low as 0.5 N with a response time under 5 ms, providing a clear signal for immediate protective stops. This direct measurement allows for precise Power and Force Limiting (PFL) compliance under ISO/TS 15066, as the sensor data correlates directly to the pressure and force limits specified for human body regions.

Joint current sensing takes a different approach by monitoring the motor's electrical current draw to infer external torque. This is a software-based, sensorless method that leverages existing hardware, resulting in zero additional hardware cost and no mechanical compliance introduced into the system. However, this strategy must filter out significant noise from motor cogging, gearbox friction, and drive-train inertia, which typically limits its sensitivity to detecting forces above 10-15 N and introduces a detection latency of 20-50 ms. This makes it a cost-effective solution for larger robots where higher payloads already exceed collaborative force limits.

The key trade-off: If your priority is achieving the lowest possible collision forces for true close-quarters collaboration with a small cobot, choose external torque sensors. If you prioritize minimizing hardware cost and complexity for a larger industrial robot operating with Speed and Separation Monitoring (SSM) as the primary safety mode, choose joint current sensing as a secondary, redundant safety layer. The decision hinges on whether the application demands the high-fidelity, low-latency data required for PFL operation or if a more economical, sensorless approach is sufficient for the risk assessment.

External torque sensors excel at sensitivity and precision because they measure force directly at the point of contact, decoupled from the motor's own inertia and friction. For example, a 6-axis force-torque (F/T) sensor mounted at the robot's wrist can detect contact forces as low as 0.5 N, enabling a robot to react to a gentle touch without causing injury. This direct measurement allows for a much lower 'trigger threshold,' making it the preferred technology for achieving the stringent Power and Force Limiting (PFL) requirements for transient and quasi-static contacts defined in ISO/TS 15066, especially in sensitive assembly or direct human-robot handover tasks.

Joint current sensing takes a fundamentally different approach by using the motor's existing current draw as a proxy for external torque. This is a software-defined, sensorless strategy that eliminates the cost, cabling, and mechanical compliance of a physical sensor. However, the signal must be filtered to remove noise from the motor's own operation and gearbox friction, which introduces a critical trade-off: a necessary time delay. This filtering typically results in a 10-20 ms latency in collision detection and a higher minimum force threshold, often in the range of 10-30 N, making it less suitable for applications requiring the most sensitive touch but highly effective for preventing crushing injuries in logistics or palletizing where a heavier payload is the norm.

The key trade-off: If your priority is achieving the lowest possible collision force for true close-quarters collaboration and you can absorb a $3,000-$5,000 per-sensor cost, choose external torque sensors. If you prioritize a robust, cost-effective safety layer for a standard industrial arm operating at reduced speeds, and can accept a slightly higher collision force threshold, choose joint current sensing. The decision ultimately hinges on whether your risk assessment requires the sensitivity of a direct measurement or the simplicity of a sensorless model.