Open Cut » Health and Safety
There is currently limited research available to prove the accuracy and reliability of Proximity Detection Systems (PDS) in production environments or to provide operations with a realistic and manageable framework to determine if a particular PDS unit is appropriate for their conditions and needs. Reliability and functionality claims made by suppliers are not easily verifiable by the end user, at least not relative to any known framework, methodology, or standard. Additionally, there are many PDS units and multiple sensing technology categories used such as radio frequency, infrared, radar, ultrasonic, LIDAR, and combinations thereof. Sites can be further confounded when selecting a PDS unit because little is known about the actual strengths and weaknesses of the various sensing technologies or what is effective/ineffective relative to operating conditions.
This project has made an attempt at developing a PDS Validation Framework that is scientifically rigorous yet practically achievable for site to implement. The project, which involved an initial investigation into the fundamental problems and challenges of validating such systems, proposed a staged 2-tier approach to PDS validation:
- Tier 1 involves validating the PDS's Object Detection capability against a set of environmental and vehicle speed variables;
- Tier 2 covers the validation of PDS's L8 and L9 capability (see PR5A L1 - L9 hierarchy of controls) in limited choreographed test scenarios.
The basis of the tiered approach is to tackle the challenge in bite-sized chunks. These chunks of tests are highly focussed and should provide clear and conclusive results on which part of the PDS unit's performance may be deficient. It is the intent that any poor performance demonstrated within either Tiers of testing will invalidate a PDS unit. If a PDS unit is inherently unsuitable due to susceptibility to any of the two suggested Root Causes, by design the tiered approach should require minimal time and resources to quickly but convincingly demonstrate the fact. Additionally, assessing the object detection layer (Tier 1) independent of the intelligence layer (Tier 2) and vice-versa creates less complicated and more manageable tests. The various variables of the operating environment and operating parameters that could affect the PDS unit's operation would not be covered in Tier 2, as these variables (if they are significant factors at all) would have affected the outcome of Tier 2 tests through their effects on the object detection capability, which would already have been discovered in Tier 1. Thus Tier 2 testing can focus on interaction scenarios and the evaluation of the PDS unit's decision-making instead of being distracted and confounded by the inclusion of operating environment variables.
A framework for Learning and Knowledge Capture driven by all stakeholders is extremely important as this will ensure that performance failures in the application environment (if and when they do occur) is converted into critical information that helps drive the development of the next generation of improved and more robust products while simultaneously reducing the set of 'unknown unknowns' of the operating environment.
The supported next phase involves translating and finalising the proposed test procedures in this body of work into a safely executable field test program that preserves the original rigour of the methodology. The next phase will also involve field verification of the proposed test program, including logistics of setting up the tests and executing them. Gaps and weaknesses of the current methodology are to be identified, and the methodology is expected to be fine-tuned and improved as a result.