IP Licensing

DeepSLAM is a software system based on deep neural networks which enables camera pose determination and 3D environment mapping in real time using single-view, unlabeled camera images.

The background

Visual Simultaneous localisation and mapping otherwise known as SLAM, technique is one of the most important research areas in machine vision and robotics with a broad range of vertical applications spanning from autonomous vehicles to virtual reality. SLAM systems can estimate the robot pose and environment maps and their uses also include robots for inspection, such as visual inspection and assessment of industrial equipment and infrastructures in harsh environments. 

The majority of visual SLAM techniques are based on vision geometry and optimisation algorithms, and use stereo images. These systems cannot learn automatically from raw images or benefit from continuously increased datasets. There are some visual SLAM techniques which are based on deep neural networks. However, these systems are trained on defined, labelled data sets. Labelling large amounts of data is difficult and expensive, which limits the potential application scenarios. Furthermore, visual SLAM systems have typically suffered from reduced accuracy in challenging conditions such as low light. These limitations have significantly impeded the adoption of SLAM technologies in key application industries.

Researchers at The University of Essex made a significant breakthrough recently. The accuracy and robustness of their system outperforms the existing SLAM systems. It has the potential to make further improvement as more training data becomes available.

The technology

A team at the University of Essex have developed a novel monocular SLAM approach, based on deep neural networks, which generates a pose trajectory, depth map, and 3D point cloud simultaneously. This patented system is trained using unsupervised deep learning, allowing the system to benefit from continuously increased data sets to update and improve its performance in real time as image data is received from each new environment. It is therefore not restricted to the finite, pre-set environment of the training data.

The essential elements of the invention are the system architecture (Figure 1), the system training scheme (Figure 2), and the computer code.

The system architecture includes two main deep neural networks: Mapping-Net and Tracking-Net. The system input is a monocular image sequence and for each image, the Mapping-Net estimates the depth and the Tracking-Net estimates the pose. Additionally, the system includes a loop detection network (Loop-Net) that assesses spatial and temporal image losses to reduce the accumulated drift in pose estimation and fine-tune the pose accuracy, thus continuously re-training the system in each new environmental context. 

The benefits

• Unsupervised learning reduces the system’s reliance on annotated ground-truth data sets, decreasing the demand for labelled data which is limited in its availability and costly and labour-intensive to generate
• Application environments are not restricted to those with annotated ground-truth data sets, greatly expanding the range of applications when compared to supervised SLAM system
• Performance is continuously improved in each new environment
• Enhanced performance in challenging environments such as low light, using just a single, cheap camera 

Applications

This technology could be employed in the following areas:

• Self driving cars
• Unmanned aerial vehicles
• Autonomous underwater vehicles
• Robotics
• Space exploration vehicles and robots
• Healthcare (guided and robotic surgeries)
• Augmented and Virtual Reality

Patent status

• UK patent application filed March 2018
• PCT patent application filed March 2019

Researchers from the School of Biological Sciences developed the DNAzymes technology to help men with advanced (stage 4 or 3) prostate cancer that require effective treatment in circumstances where current therapies would be expected to fail succeed by providing novel method to inhibit therapy resistant disease 

The technology

DNAzymes are short single stranded pieces of DNA with catalytic activity. DNAzymes do not require cellular machinery to cleave their target RNA. Ribonucleic acid or RNA is a polymeric molecule essential in various biological roles in coding, decoding, regulation and expression of genes.

DNAzymes have binding arms allowing selectivity for their chosen target and a catalytic core that cleaves the RNA. DNAzymes are optimised using a number of parameters to ensure cleavage efficiency.

The problem

Prostate cancer is the most common cancer diagnosis in men and is the third leading cause of cancer-related death in the UK. Current therapeutic procedures involve surgery often coupled with radiation therapy and in advanced cancerous stages, hormone therapy and chemotherapy.

Prostate cancer is associated with alterations in Androgen Receptor functions. The Androgen Receptor signalling pathway is a fundamental process in the growth of many cancers and in particular prostate cancer. Current methods of treating this type of cancer take advantage of this androgen dependence by disrupting the signalling pathway to prevent tumour growth.

Although initially successful in the majority of patients, these therapies invariably fail and the tumours progress to an aggressive therapy resistant stage, termed castrate resistant prostate cancer. Importantly, Androgen Receptor signalling continues to drive tumour growth, due to e.g. receptor mutations or alternative splicing events, and therefore remains a therapeutic target for the disease.

The context

There were around 46,700 new cases of prostate cancer in the UK in 2014, that’s 130 cases diagnosed every day making prostate cancer the second most common cancer in the UK. There were 11,300 deaths caused by prostate cancer in the UK in 2014, 92,300 deaths in Europe and 307,000 worldwide in 2012. The global prostate cancer therapeutics market was valued between $5 billion to $9 billion in 2016 according to different reports.

The solution

DNAzymes have been designed to target the Androgen Receptor and to be active in circumstances where current therapies would be predicted to fail (e.g. Androgen Receptor splice variants or mutations). DNAzymes have been identified that successfully cleave their target mRNA with high efficiency.

(a) Schematic summarising how DNAzymes work.
(b) Cleavage reactions were performed and % cleavage of the target RNA quantified.

Our lead DNAzymes show more than 98% activity in cleaving target RNA. Cleavage of target RNA leads to down-regulation of androgen receptor protein levels, which in turn reduces prostate cancer proliferation.

The Mission

Improving the quality of life for people with visual field loss by using augmented reality to increase visual-awareness, facilitate independence and enable safer mobility.

The problem

Visual field loss is a common consequence of a number of conditions, including stroke, tumour, trauma, and eye diseases such as retinitis pigmentosa and glaucoma. It has a significant impact on quality of life, reducing mobility and independence, creating difficulties locating and interacting with objects, and making those affected / maleficiaries vulnerable to hazards such as traffic and obstacles.  Every year, 152 000 people in the UK suffer from a stroke, and there are 13.7 million first-time cases across the globe each year (Stroke Association, 2016). Loss of vision is one of the most common and most debilitating consequences of stroke, second only to weakness in the arms or legs. Unfortunately, in contrast to the intensive therapies available for stroke survivors with damage to the motor cortex, cortical damage in the visual cortex is considered permanent. The visual field describes everything that is seen when looking straight ahead and visual field loss is when a part of that view is impaired. In this kind of visual impairment, the damage is often permanent. People with visual field loss experience significant mood disorders, loss of independence, impaired mobility and an increase of slips, trips and falls, yet care for people with visual field loss is extremely limited.

The solution

Researchers at the University of Essex are developing an assistive technology for people with visual field loss.  The software platform built for use within Augmented Reality headsets will provides the user a live stream and cues about content in their blind-field. The software will identify the location of each user’s missing field of view and will allow them to choose where this is displayed within their Augmented Reality device.

Whilst still under-going user testing with the target audience in natural environments, initial research has provided encouraging results.  In controlled laboratory tasks, having the missing information displayed in the sighted field, has improved awareness for targets of the blind-field.

Benefits

Our aim is to help people with visual field loss to be more aware of their immediate surroundings which will allow them the confidence to navigate their environment more safely.  We will use readily-available consumer headsets, rather than bespoke solutions that will dramatically reduce the cost to the individual.

Benefits include an affordable solution that promotes a greater sense of independence, the ability to carry out everyday tasks safely, improved mental wellbeing and reduced risk of isolation. In addition, linked to the above a secondary impact will be fewer accidents and therefore a reduced burden on friends and family, and the NHS.

Patent status

• UK patent application filed July 2020

Discover more about AReye.

The technology

This technology is based on a high resolution MS/MS technique where the phosphorylation pattern of certain residues within the C-terminal part of the tumour suppressor protein Scribble will enable a clinician to determine the best treatment option within triple negative breast cancer patients. The phosphorylation pattern on a cluster of sites in the C-terminal region of scribble dramatically changes when cancer cells become more invasive.

The problem

The use of chemotherapy as treatment options for breast cancer is utilised within the population of breast cancer sufferers with tumours that are negative for receptors to oestrogen, progesterone and HER2. Upon biopsy the cancer's receptors are identified to identify the best course of treatment. Within the triple negative population of breast cancer patients, chemotherapy is offered as the treatment. Triple negative breast cancer is seen as a more invasive cancer with poor prognosis.

The climate/arena

Breast cancer will affect nearly 232,000 women in the USA over the next year and 15-20%of cases will be triple negative tumours. The figures for the UK are 50,000 cases per year and the rest of Europe including Belgium, Denmark and France have higher incidence rates than the UK.

The solution

The technology described here is able to distinguish between highly invasive and less invasive metastatic state of cells. Loss of phosphorylation appears to be an early marker for invasion, metastatic propensity and ultimately poor prognosis. Whereas phosphorylation of the sites indicates a less invasive and less metastatic state of the cells. The current technology is mass spectrometry based and there is potential for an ELISA based method to be developed.

Patent status

Patent issued in the USA in April 2016

The technology

Heat-shock transcription factors (HSFs) are known to be key regulators of expression of heat shock proteins in eukaryotes. Individual HSFs have unique functions in response to environmental stress. They are known for many plant species including corn, rice, soybean, wheat, barley, potato, tomato and Arabidopsis. Prof Phil Mullineaux and Dr Ulrike Bechtold have now demonstrated that HSF3 acts beyond its previously associated functions, and have found that transgenic plants over-expressing HSF3 show benefits to water productivity and pathogen resistance.

The solution

Expression of HSF3 in transgenic Arabidopsis has shown significantly enhanced plant performance under drying conditions – leading to increased seed yield, improved water use and better recovery from severe drought stress. Moreover, plant productivity is not only improved under drying conditions but also under normal growing conditions – plants over-expressing HSF3 produced 2.5x the seed weight of normal plants even under well watered conditions. This effect was confirmed in three different Arabidopsis accessions, whereas hsf1/3 double null mutant plants showed reduced water productivity. The overall biomass was not affected, but redistribution from vegetative biomass in favour of seed biomass was observed in the HSF3-over expressing plants.

Patent status

• USA - 8445747 B2 - granted.
• Brazil - filed.

Software and code for image calibration, aligning images, and generating NDVI imagery for immediate use that uses off the shelf components.

The background

Normalised Difference Vegetation Index (NDVI) is a tool to easily and quickly assess plant and crop health.

A normal, healthy plant will have higher amounts of chlorophyll pigments, which absorb most blue and red wavelengths. Green light is less readily absorbed with a proportion transmitted and reflected from the plant, which is why they appear green to our eyes. With the green visible light, plants also reflect Near-Infrared (NIR) light. This type of light, which is invisible to the human eye, is also not actively used for the photosynthesis process and thus is not affected by chlorophyll pigment content.

When the chlorophyll content of a plant decreases, due to stress such as drought or low nutrient availability, the amount of absorbed red light decreases. This in turn increases the reflectance of red light. Additionally, when a plant becomes dehydrated or stressed, the spongy layer of the plant collapses and results in less reflected NIR light. In both cases, there is an increase in the amount of reflected red light compared to the amount of reflected NIR light.

NDVI, simply put, is a calculation of vegetation or crop health. Mathematically comparing red and NIR light signals can help differentiate plant from non-plant and healthy plant from sick plant.

The technology

The invention is an NDVI imaging and image processing system which runs on a Raspberry Pi, using off-the-shelf components and a methodology for highly accurate image calibration by applying readily available affordable reference standards.

This allows accurate calibration of the system and accurate NDVI imaging at a fraction of the price of other commercial systems, while providing modularity and customisation. Calibration is essential to obtain biologically relevant images across a wide range of light conditions and provides good results that can be compared across field sites, seasons, etc.

Currently available solutions require reflectance standards with uniform, well-calibrated spectral properties, and an optical filter that removes all light outside of the desired spectrum. These solutions, while highly accurate, can be costly and difficult to readily obtain.

The benefits

The new system has the following advantages over commercially available systems:

• Software and code for image calibration, aligning images, and generating NDVI imagery for immediate use, using off the shelf components and based on the University of Essex developed methodology
• Significantly lower cost to purchase and produce
• The Raspberry Pi platform is inherently customisable, therefore the system can be further tailored by the user to fit their requirements (e.g. add WiFi or GPS, combine with data-logging systems, add other devices or capabilities, etc)
• Easy to scale from greenhouse measurements of a few individuals, to aerial imagery of entire crops

Patent status

Patent application submitted.

In 2018 we signed an exclusive license with Cerium Visual Technologies Ltd, giving the optical manufacturer rights to our academic expertise.

The technology

The intellectual property license agreement that gives Cerium access to the university’s know-how has enabled the further development of a potentially life changing piece of technology.

The Intuitive Colorimeter ‘Curve’, developed in partnership with Professor Arnold Wilkins from our Department of Psychology can accurately and quickly prescribe tinted lenses to help some individuals with visual stress-related reading problems and associated medical conditions.

The new Intuitive Colorimeter Curve can now – for the first time - provide a fully digital, more streamlined assessment.

The solution

Having launched the ‘Curve’ in the UK in January 2018, Cerium Visual Technologies plan to expand and further build on this presence internationally.

Visual stress is thought to affect up to 20% of the general population. The 'Curve' reduces symptoms of visual stress using filters of a particular colour. The colour is individual and prescribed to patients as tinted spectacle lenses.