Current practice for preventive maintenance inspections of rolling stock over Indian Railways network and in many other parts of the world is largely based on manual inspection which is either trackside or pit examination of stock in stationary or slow moving condition. Visual inspections are performed by trained manpower either in a pit or trackside location but remains dependent on individual judgment. Besides, the stationary inspection deprives the maintenance staff of significant information of dynamic behavior of the vehicle. Automated inspection by machine vision based systems has the potential to overcome these limitations of human inspection. The systems can be placed closer to the track or between the rails where it may be considered unsafe for a human to be positioned when a train passes. A number of defects may be automatically detected using a combination of visual sensors and automated analytics; these may include side view defects on missing and broken components related to axle, wheel disc, wagon doors, springs, shock absorbers, wheels, brake pads; underframe defects on malfunctioning electrical components, cracked components, oil leakage, hanging parts etc., and top view defects including deformed side walls and hanging doors of wagons. Current market solutions are limited in their coverage and performance. As a part of this project, Rail Vision will adapt its existing highly successful technology for track inspection to work for train inspection. It will focus on adapting the electro-mechanical interfaces, sensor positioning and sensor triggering modules for high quality data acquisition of moving trains. It will collaborate with two key customers in India to adapt the technology and is expected to receive future funding from them and other sources to manufacture and trial out the technology at their end. It will also apply for patents in this area to cover its intellectual property. Once the technology is trialled and tested, Rail Vision shall be able to export this technology to a range of international markets.

This 18 months industrial research project will develop a SafetyNet device for interpretation of x-ray images from security scanners for carry-on, hold luggage and parcels using automated image interpretation with deep learning technology. The system will automatically determine the level of threat posed by the object under the scanner, and provide a safety net around the human operator decision making. The device output will be integrated with existing scanners and used to alert the operator to the presence of threat items. The device will be tested on dual energy x-ray equipment in laboratory and at customer sites. The device will have universal appeal, integrating to a number of scanner models, making it a perfect option for retrofitting and new product line integration. The underlying algorithms and detection technology will be based around the use of adaptive self-learning neural networks which use underlying principles of deep learning. Such systems will learn from millions of images scanned and tested by human operators in an online setting. Once trained these will be applied for threat detection to assist human decision making. As more data is scanned, these systems will have larger archives to learn from, and will continue to grow their intelligence over time. The systems are expected to markedly improve performance in terms of increased quality of threat detection, higher screening throughput and reduced false alarms and resulting delays. SafetyNet will be offered to potential customers as a combined hardware/software offering with a cloud backend to which the device will communicate data on system and operators, which can be used subsequently for delivering data services to customers to report on reducing the risk of visual fatigue and inadequate staff training. The target markets for such system will include stadiums, arenas, border control, ports, aviation, commercial head offices, data centres and defence applications.

.

Year round UK glasshouse production of tomatoes for supermarkets is highly intensive. Although this production is highly controlled it is still difficult to accurately predict picking yields which has an impact on the food supply chain. This inaccuracy requires businesses to continually react to this inefficacy. Over-prediction results in costly imports, whilst under-prediction incurs financial losses from the disposal of surplus fruit. There is considerable potential to reduce these losses and increase the proportion of UK sales by improving the accuracy of yield forecasts. We will develop an imaging system, TomVision, and mathematical models, PredictTomPro, to more accurately predict weekly yields and deliver significant savings in import costs and waste. Our aim is to predict weekly harvests to 10% of actual, that will generate £30K/ha extra income p.a. for producers and for the developers the anticipated sales of these tools of £1.3M/£11.3M/£26M (UK/EU/W) after 5 years.

Train accidents are often due to worn bogie parts (wheels, axles, brakes & bearings) overheating, breaking, & derailing the wagon affected. To identify this heat, Hot Axle Box & Hot Wheel (HABD/HWD) detection units are mounted at the track side. HABD/HWD use infrared (IR) sensors to makes point based measurements of passing bogie IR radiation, which is converted into electric voltages & displayed as absolute temperatures. There are limitations with HABD/HWD, such as sampling of incorrect target points giving inaccurate information, as well as detrimental effects from solar radiation. ~10% of heat derailments occur despite the train having previously passed a HABD/HWD – likely an underestimate, as Germany, where most EU HABD/HWD are installed, has the highest number of heat-related derailments. The opportunity is for Rail Vision to develop a system to detect heat from bogies via a £582,405, 24-month development of prototype project. This low-cost, undercarriage-mounted digital thermopile array module (DTAM), coupled to an optical camera, creates a continuous bogie monitoring system (CBMS). The CBMS tracks & records heat, merged with optical data from the bogie to provide more accurate information than HABD/HWD, & enabling remote, real-time monitoring (RTM) of the bogie. The CBMS will detect a gradient of temperature across a cross-section of the component being monitored – to do this, RV will generate bespoke algorithms, data communication protocols & software platforms to collect, track & monitor data, & alert. CBMS RTM enables more accurate evaluation of temperature anomalies than isolated IR HABD/HWD sampling – for customers, this will result in greater safety of their rolling stock, and maintenance cost improvements. The CBMS supersedes HABD/HWD by monitoring temperature across cross-sections of bogies vs. isolated areas, allowing the user to accurately pinpoint heat anomalies via merging with optical imaging, & tracking & building a history of bogie performance

The project aims to engineer a high impact advance in crop stress monitoring through the development of a novel, flexible multi-sensor imaging system (HD video, IR, Flourescence) for application on a mobile platform (manual or robotic) to automatically detect stresses in crops initially for tomatoes in a protected environment. It is aimed to support growers to improve yields whilst also reducing environmental impact and labour intensity. The project builds on recent research into the use of multi-sensors for the detection of crop stresses and the application of advanced image processing analytics to provider an automated crop stress monitoring solution. Trial units will be developed and outputs displayed.

This project aims to prove the feasibility of producing a high impact advance in surveying of the railway network through the development of a novel device capable of high speed asset monitoring and automated asset identification for the railways. It is aimed to support the work of Network Rail and their sub-contractors who require detailed asset maps of the rail infrastructure. The project builds on recently patented IP from Oxford University Mobile Robotics Group. The combination of Laser Scanning and HD camera hardware will combine with satellite navigation systems to create a 3D topometrically correct asset map of the rail network which is automatically analysed with the latest visual analytics techniques. Trial units will be developed and outputs displayed.