Construct - Issue 49

and there’s usually only one point of entry,” says Adam.

The first task is to close the section of road or footpath and carry out a visual assessment to identify what the defect is. This is usually done with endoscopic cameras, but these give limited information and require putting people in proximity to the defect. “The next port of call is to decide if we need to do a full dimensional survey and condition report,” explains Adam. “At the moment, that means sending a highly trained operative into a challenging environment to gather information. It could be a confined space that is dark and dirty, partially damaged or collapsed, with uneven brick or rubble. From a safety and risk point of view, we want to avoid sending someone into that environment if we possibly can.” “We can take pictures and collect dimensional information with enough accuracy to give us the critical information quickly” This is where the robotic quadruped comes in. The device being trialled – dubbed ‘Robodog’ by the team – was initially developed for search and rescue, so it is extremely robust and can fit into very small spaces. It walks across uneven surfaces easily and can collect all the data the team needs to assess a void and its condition. “The Robodog is fitted with a 180-degree field of view high resolution camera and built-in Light Detection and Ranging (LIDAR), so we can take pictures and collect dimensional information with enough accuracy to give us the critical information quickly,” says Adam.

“If it is operating in an area with poor lighting, it can even use the LIDAR to navigate.”

The robot can be controlled from any smartphone via the device’s built-in Wi-Fi or fitted to a video-game type controller. “It’s super intuitive,” says Adam. “The simplicity is one of the things that makes it very attractive.” The device’s AI and machine learning capability mean it can be ‘taught’ – for example to recognise or avoid certain objects – and it can be fitted with attachments, including a robotic arm to collect samples. “When we do a traditional visual inspection, we have to get close to the collapsed area,” explains Adam. “We do a plan of work and a risk assessment, but there is always an element of engineering judgement when you’re trying to assess how close is safe. By using a robotic device to carry out that assessment, we could completely eliminate that risk profile.” The team is spending three months trialling the device, putting it through its paces in all the different environments and scenarios it could encounter before deciding whether to invest in one. “I think it has huge potential – not just for our team but for all teams across the business,” adds Adam.

19