Following an LHD operator fatality at an underground mine site, a mine is mandated to monitor and introduce measures to improve seat belt usage among its employees operating heavy mobile equipment. There are existing systems to warn an operator of an unbuckled seat belt while operating, however this system is confined to the cab of the vehicle. The mine needs this data delivered to analysts on the surface where it can be used in the detection of events and alerts sent to key stakeholders in the operations and safety departments. Due to the working environment the solution needed to be rugged, and be able to function in a space of intermittent network connectivity.
To solve this problem the site came to Symboticware due to its previous experience working with them to collect data from the LHDs, recognizing their previous success in operating in the mining environment. Symboticware worked with the provider of the onboard seat belt monitoring system and integrated its SymBot solution with it so that the SymBot could collect the buckled/unbuckled status of the seat belt. In addition to this, the SymBot collected contextual data such as the LHD’s ground speed and remote switch status from other devices on the machine.
Through Symboticware’s patented store-and-forward technology the datasets were comprehensive and it was all transferred over the mine site’s existing WiFi infrastructure in a FIFO manner each time the LHD came under wireless coverage. From this point, this information was used by Symboticware to develop two separate alerts to identify instances of dangerous seat belt usage.
The first alert was in the instance of a clear seat belt violation, which was defined as: seat belt unbuckled, positive ground speed, and remote switch not engaged. With the remote switch engaged the operator was presumed to be outside of the vehicle, meaning that positive ground speed with the seat belt unbuckled was an appropriate state.
Below is an example of this state: the seat belt switch status is shown in blue, the remote switch status in red, and ground speed is in grey. In this instance the machine was at a stop, the seat belt unbuckled, the remote switch off, and then the vehicle was moved briefly. It was later learned that the machine was parked in an area obstructing the path of another piece of equipment. The operator got into the vehicle to move it for them and failed to buckle their seat belt for what they described as “only a minute” of operation.
Alerts of this state were presented in a web-based dashboard, as well as through an immediate email alert once the state was detected. The operator was disciplined for this seat belt misuse.
The second alert type came later. Initially the seat belt violation type shown above was sufficient, but it was later determined that operators were not only committing seat belt violations, but they had learned a means around the in-cab siren deterrence measures that had been implemented. To do this they would simply buckle the seat belt behind their back, and operate without being strapped in. The stakeholders at the mine returned to Symboticware with this information, and together they were able to find a means to catch this behaviour which they called: seat belt avoidance.
To catch instances of this behaviour Symboticware reviewed the data and found instances where the seat belt remained buckled during periods of remote LHD operation. As the operator needs to depart from the asset to operate it in remote mode—and it was considered unlikely that they slipped out from under the seat belt rather than unbuckle it—it was a clear indicator that they were engaged in circumvention of the seat belt use compliance system.
With this lesson learned, a new alert for the following status was added to the seat belt violation detection: seat belt buckled, remote switch on, and ground speed positive. Below is an example of this alert type, each time the seat belt (in blue) and remote switch (in red) are buckled and on, an instance of this seat belt avoidance occurs.
In under an hour the operator exited and entered the cab eight times without unbuckling their seat belt. This was a clear violation of seat belt use policy.
With these two types of alerts prepared the mine site’s management team was able to begin taking corrective actions. Their first step was to notify the operators that they were now monitoring their seat belt use and that action would be taken should it be found they were engaged in seat belt misuse. Simply knowing that they were being monitored for this type of behaviour caused an immediate improvement in adherence to seat belt use policy. Next, they began with retraining for offenders not deterred by the knowledge that they were being observed. Finally, they moved on to termination of employees that were repeat offenders presenting a liability to the organization.
Below shows the impact of these measures, summarising the week-by-week improvement in seat belt compliance. Initially in the week of July 2nd no seat belt avoidance incidents were detected, but the impact of that knowledge is evident in the next week when the overall seat belt infractions declined marginally while the seat belt avoidance numbers surged. This represents cases of operators choosing to navigate around the system, rather than adhering to the seat belt use rules.
The week of July 16th is when operators were informed that both violations and avoidance events were being monitored. Following this there is a steady decline in incidents until zero events were detected over the course of four weeks, at which point the project was deemed a success. Prior to this there were a few incidents of seat belt violations and alert avoidance actions, however these were increasingly sparse.
Over the course of the next year the infractions were monitored. Occasional incidents occurred, but there was an overall 99.9% reduction in seat belt use infractions from the time the project began and the system was implemented. Newly hired operators are warned on hire about the zero tolerance policy onsite towards seat belt misuse, and alerts are a result of willful negligence rather than a lack of training.