Einstein said, “Everybody is a genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is stupid.” That’s what we are doing with AI. Focussing more on what it cannot do as opposed to what it can achieve better than humans. Playing a game of Jeopardy forever, or getting it to pass the Turing test. We do not need AI to replace people, but to help people create safer workplaces and make processes more efficient, find patterns and prevent repetitive injuries et al.
Part of the problem with AI is the perception that it is a technology of the future. There are way too many movies made on dangers of AI and not enough actual case studies that have reached the masses. The most unpopular opinion about AI is that automation leads to job-cuts. In developing countries with high population density and high unemployment rates, this impression deters some enterprises from even exploring AI-based solutions.
AI To Help Achieve Zero Accidents
According to International Labour Organisation, work-related injuries and deaths cost the world almost $3 Tn, this is approximately the GDP of India. HSE is one field where AI for automated risk identification could save several lives, every single day, all through the year. If you think about it, CCTV cameras are as ubiquitous as they are useless. Only used in retrospect after something goes wrong and even then the process is cumbersome and time-taking.
Computer Vision is a field of Artificial Intelligence that can analyse live feed from CCTV cameras with deep learning and give actionable insights, real time. Computer vision solutions aim to achieve zero accidents in manufacturing premises. In most cases, a computer vision solution can be plugged to an existing CCTV network without needing an infrastructure overhaul. The solutions often come out of the box and and can be deployed in a very short timeline.
Using all the cameras in a premise, a computer vision solution will create a virtual map to analyse the total operational area, mark hazard areas, define non-compliance and configure the type of alerts to be raised. The alerts can vary from sms or email sent to the right authority, announcement on PA systems or buzzers at a particular point.
The hierarchy of authorities is also fed to the AI solution for taking action in case of a critical alert raised. Real-time notifications are raised every time there is a non-compliance. Dashboards providing insights can be accessed on app or on web. These help the enterprise to take steps to rectify a recurring problem.
Some of the real scenarios in which AI makes a huge difference by working with existing security protocols providing tangible results, every time.
Traditional fire detection systems use smoke and thermal detectors which are unreliable and slow. The response time is around 3-5 minutes, often the fire has spread to a large area and is difficult to contain or douse.
An AI solution can detect fire within 10-15 seconds from a distance of about 50 metres. It can also give the exact location of fire and send alarms to concerned authorities, escalating if no response is received. It can be integrated to the PA system to immediately announce the exact location of the fire and the exit route to be taken.
AI-powered solution can also identify that the number of people that entered the building is less than the number of people that exited the building. Using the head-count it can guide fire-fighters to the right floor in case of a multi-storeyed building, or the right area in case of a large warehouse or factory, saving critical time.
Mechanical Equipment Used For The Movement (MHE) Safety
Pedestrians face the greatest danger in a manufacturing premise, warehouse, yard depots etc by walking into the pathway of MHE. A forklift driver will not have a clear field of vision when the forklift is loaded, he/she will also not be able to identify near-misses because they would not have seen it.
AI can help identify and manage separate pathways for MHE and pedestrians and raise immediate alarms to prevent accidents. It can create a heat-map of violations to identify hazard zones that help enterprises create an SOP that reduces accidents in those areas.
Workplaces, products and systems are ergonomically designed to work at optimum by when used in a pre-defined manner. Ergonomics aims to create safe, comfortable and productive workspaces by bringing human abilities and limitations into the design of a workspace, including the individual’s body size, strength, skill, speed, sensory abilities (vision, hearing), and even attitudes.
A faulty pose is the most difficult trauma but is often passes undetected by human eye. This is the reason lower back pain is the most common workplace injury. AI can monitor the pose for work requiring the use of force exerted by a person to lift, lower, push, pull, carry or otherwise move, hold or restrain. Compliant pose for manual handling work is defined and monitored.
AI can identify patterns where a wrong pose is repeated, leading to injury. This helps the workers by alerting them before they end up being injured. Saving not only the people from injury, but also in the long run, building positive image of the enterprise by being a humane place that cares for its employees and is very safe to work in.
Reducing Workplace Accidents By Ensuring PPE Compliance
AI can identify workers not complying with the correct PPE, for example – helmets, vests, boots, spectacles etc raising alarms in real time for even every non-compliance. This leads to an immediate rectification of the problem and saves a future accident, amputation or injury. Face-recognition technology can be used to limit specific people in hazardous zones, raising an alarm when a breach occurs.
Construction sites that employ workers for working at height can deploy an AI solution to ensure that safety harness is always securely fastened. Using live feed, real-time alarms are raised if a persons harness is unfastened or if there is too much crowding at a particular scaffolding. Computer vision can also play a particular message like ‘wear helmet’ if someone is entering a restricted area without the required PPE.
Eventually, the purpose of AI is not to replace humans, but to help us build a safer, more efficient work environment. Not exactly Artificial Intelligence, but Augmented Intelligence. The reason Picasso is irreplaceable because he was a person, made of several complex things – biological, psychological, emotional, circumstantial and the impossible to replicate – ‘stroke of genius’.
Quantum AI is still in the concept phase, AI-based on binary coding AI is considered phenomenal when it can achieve the cognitive intelligence of a two-year-old. More media coverage is given to AI when created music, painting or trailers than is given to the boring but greatly affecting life-saving AI that works for HSE.
Accident numbers do not have an identity, it puts people together as faceless, non-flamboyant, blue-collared workers – replaceable. But for an AI solution, each worker is different, identifiable. AI is able to focus each of these people and definitely save them from the next accident, the next fall, the next amputation. Nothing escapes the AI-powered CCTV camera.
Enterprises of all sizes can deploy AI-powered solutions, specifically for HSE.
Because when you think of the cost of one human life, it is actually, priceless.