" Headlight AI have developed ‘Telesto’ as a LiDAR survey platform, to accurately record the intrados and dimensions of tunnels in a semi turbulent flow. Wessex Water were instrumental in bringing forward the concept of Telesto and to date we have commissioned over 8Km of tunnel modelling via the system and this will be used as a baseline, allowing further iterative surveys to monitor the condition of the structure. The autonomous platform reduces the need for us to deploy surveyors into our tunnels and this is a major health and safety benefit. "
" For a while now we have been looking for an enhanced surveying technique that goes quite a bit further than existing laser profiling options that currently lead the way but stops short of full blown survey grade LiDAR. Such solutions are not well suited for the harsh environments, especially non man entry applications. Headlight AI’s solution ticks that box bringing clever machine learning technology as they are the experts in getting the very best results from what the “off the shelf” LiDAR scanner churns out. "
Robots today are smart. They can make sense of complex visual or auditory patterns, avoid obstacles, and move around autonomously. But when it comes to harsh conditions they are often fooled or fail completely. This is a problem that already affects robots in industrial applications, and it is preventing the development of new autonomous systems that could save lives across a wide range of applications, including infrastructure monitoring, transportation, and emergency services.
Conditions where sensors usually fail.
For example, we are developing software that allows sensors to accurately collect data in harsh environments. We are starting within challenging infrastructure environments across the water and rail networks. This will help to:
When we enter a new environment, our senses adapt. We think machines should do the same. Our patent-pending technology allows machines to adapt to their environment. This optimises both data collection and quality. We can help you get the most from your existing sensor network and integrate these sensors with our AI software. Alternatively, we can help you select new sensor systems that are compatible with our AI software, to ensure you can map any unknown environment.
Running AI software locally (on the edge) is more complex than across a network (e.g. in the cloud). You need energy efficient algorithms, especially for light weight machines (e.g. drones). We are developing efficient deep learning algorithms that can run on low power devices (e.g. FPGAs). This extends the operational life of machines that are reliant on batteries, and allow them to keep making intelligent decisions. For example, to save power in ideal conditions, to report a defect on a safety critical asset, and to adapt to their environmental conditions.
We provide automated reporting software for infrastructure monitoring. This solution can
help you to monitor your safety critical assets and provide optimised maintenance
meaning the right people can get to the right place at the right time.
Our current software can operate in complex locations; deep underground, in confined
and across high-speed transport networks (e.g. rail).
We can enable you to see in the dark or in low visibility, build digital catalogues of your assets in real-time, and scale solutions from edge devices to the cloud. In the cloud, we can provide further tools to post process your data and conduct further analysis and reporting.