A Path for Realistic Human-Robot Collaboration—Part 2
By Marek Wartenberg, Senior Engineer, Robot Dynamics and Controls, Veo Robotics
In Part 1 of A Path for Human-Robot Collaboration we discussed how current manufacturing trends toward mass customization and faster product cycles mean that manufacturers can't amortize the costs of fully automated workcells; they need human production workers. The rapid surge in popularity of power and force limited (PFL) robots has shown that collaborative applications can increase productivity, provide faster fault recovery, and increase unit production rates. But although PFL robots can perform well for certain applications, they are ultimately too weak and slow for most durable goods manufacturing, and lose their advantage if equipped with a dangerous payload or end effector. Instead, a vision-based implementation of SSM provides a way to overcome these limitations, making large industrial robots aware of humans (see Part 1 for further detail).
In Part 2, we will discuss how the stopping characteristics of industrial robots affect performance of Speed and Separation Monitoring (SSM), and how increased performance, characterization, and reporting of stopping data by robot manufacturers can improve the functionality of SSM.
The Protective Separation Distance
The PSD is calculated between the robot and any area or volume that a person could potentially occupy. A safety-rated Protective Stop is initiated if the PSD is violated. The robot can automatically restart and continue its trajectory once the operator moves away and re-establishes the PSD. The PSD calculation is defined in ISO/TS 15066 as:
Sp(t0) = Sh + Sr + Ss + C + Zd + Zr
Where:
Sp(t0) is the protective separation distance at t0;
t0 is the present or current time;
Sh is the distance attributable to the operator’s change in location;
Sr is the distance attributable to the robot system’s reaction time;
Ss is the robot system’s stopping distance;
C is the intrusion distance, defined by ISO 13855 as the distance a human body part can intrude into the sensing field before it is detected;
Zd is the uncertainty of the position of the operator in the collaborative workspace as measured by the presence sensing system’s measurement tolerance;
Zr is the uncertainty of the position of the robot system, resulting from the accuracy of the robot position measurement system.
In practice, the result of this equation is dominated by two components: Ss, the stopping distance of the robot, and Sh, the distance a human could move toward the robot while it is implementing a Protective Stop. Manufacturing engineers and system integrators will often calculate a worst case PSD based on the max speed and load of the robot and install a laser scanner or light curtain at a distance larger than that max calculated value. This static calculation might not take into account the application at all and often leads to wasted factory floor space and limited access to the robot for collaborative tasks.
Figure 1. The robot performing a task at Fixture A while the operator performs a task at Fixture B. The minimum distance line is shown in green, meaning the distance between the two is larger than the PSD. When the distance falls below the PSD, the line turns red and a Protective Stop is initiated.
The alternative is what we’re doing at Veo. The Veo FreeMove™ system calculates the PSD thirty times a second based on the dynamic state of the workcell. These dynamic calculations allow a human and a robot to be in closer proximity at slower speeds, as opposed to being stuck at a fixed precalculated distance. The figure to the right shows the point cloud representation of a workcell from data captured by the FreeMove Sensors™. In the dual-fixture workcell, a robot is performing a task at one station while the operator performs their value added task at the other. The green line indicates where the minimum distance between the two is larger than the PSD.
The Data
The stopping time and distance for different dynamic states of the robot (33%, 66%, and 100% of speed, load, and extension) are provided by robot manufacturers per Annex B of ISO 10218-1; however, this data is often fragmented, hard to interpret, overly conservative, and generated by simulation. The requirement itself is interpreted differently by different robot manufacturers—some provide this data in tables while others provide it in graphs that vary in resolution quality.
Veo Robotics is serious about safety, but we also understand that more collaborative productivity is possible if stopping data more accurately reflects the true dynamics of the robot. For example, in some cases the stopping data indicates that it will take the same amount of time for the robot to stop whether it is moving at 33% speed with 33% load, 100% speed with 100% load, or any combination in between. This is likely because the worst case number for 100% speed and 100% load was propagated throughout the table. Using this number for all cases does guarantee a safe-state no matter the robot’s speed or load, but doing so also hugely limits the functionality of dynamic SSM.
Furthermore, because the governing ISO standard only states that data should be provided for those three speeds, the lowest being 33%, there is no guidance around how to extrapolate this data down to zero. The charts below show the stopping distance and stopping time for a particular robot joint provided by a robot manufacturer. The stopping distance shown on the right visually extrapolates to zero. At the extreme case of 1% speed it makes sense that the stopping distance will be much closer to zero than the stopping distance at 33% speed.
Figure 2. Stopping distance (left) and time (right) data provided by a major robot manufacturer for a joint on a common industrial robot arm.
The stopping time, however, appears to extrapolate toward 400ms as the speed nears zero. This non-zero extrapolation has a big impact on dynamic SSM. ISO 13855 states that an operator’s limb should be considered to move at 2m/s, meaning the contribution of human movement to the PSD is 0.8m for all speeds at and below 33%. If this value extrapolated closer to zero, the human and robot could safely operate at a closer distance at speeds below 33%.
Moving Forward...
Ultimately, safeguarding using dynamic SSM works well today and major manufacturers are buying the technology. But throughout the development of this technology, Veo has identified key ideas where robot manufacturers and industry participants can work together to improve the performance of all sensor-based industrial robot control. In lieu of directly improving stopping functionality, dynamic SSM can become much more efficient if, as an industry, we simply report more accurate and detailed stopping performance data.
Stay tuned for Part 3 where we will discuss controller latencies and their impact on system performance.