Case Study – (ongoing) – 5G-Victori Horizon 2020 programme partners

We could probably fill a whole website alone – in all that’s been achieved with Polaron in the course of 5G-Victori. Urban Hawk used Polaron to demonstrate a wide range of capabilities – primarily focussed on the development of edge based technologies and the use of the ultra high performance of 5G networks and their ability to swap edges (similar to moving between cell phone towers – but at ultra high speed and performance and far more localised).

Polaron enables a range of capabilities for testing the capabilities of 5G.

Polaron is a real-time simulation tool capable of ingesting live API data, GIS scale maps, and real-time sensor data such as Computer vision, Lidar, Smart Phone scans – and then traverse that data with ultra-fast agents to map and model change and predict events, empowering you and your business with greater insight and situational awareness.

This allows us to ensure we can deliver the types of capability and experiences that only 5G can enable.

Polaron as an edge based Simulation Service

In this configuration Polaron is functioning as a SaaS solution – accessible via any modern web browser.

By leveraging the other capabilities of Urban Hawk – we’ve been able to bring a huge diversity of complimentary software skills and capabilities that can help refine and develop solutions for the business needs of the 5G-Victori consortium.

Whether it’s real-time collaborative planning, to pathfinding and multi user tools allowing real-time decision making.

Through to “what if” and “wargaming out” potential solutions. Polaron uses a combination of backend and frontend web technologies to allow for easy interaction between the simulation (where the heavy lifting is done) and the frontend. This has ranged from insurance applications for Parametric Insurance, through to ingesting SLAM and Lidar scans of physical sites to allow indoor and outdoor navigation.

Each release requires refinement across the stack.

Through this programme we’ve been able to generate 200 km2 simultaneous modelled area in real-time Based on 1:1 scale 1m2 model running in real-time.

We’ve stress tested this with 10 x Concurrent editing users interfacing with a single mapped simulation (2D & 3D).

For consumption only or viewer based maps – we believe this can be served up to hundreds and thousands of users and even more with cached map retrieval.

At the last performance test – it took 112 seconds to “unpack” the entire UK Open Street Map DB Inc Error Correction.

Based on feedback from the consortium we then create 50km2 tiles.

We’ve also scaled this to the whole EU OSM Db – where it took 152 minutes to “unpack” the entire EU Open Street Map DB Inc Error Correction.

It takes 350 milliseconds to then load a 50km2 tile with all the data contained in it.

We’ve also been able to create a real-time lidar and EOS pipeline with typical time to ingest and render point cloud from a typical Velodyne Lidar (determined by read/write of Motherboard / CPU / GPU). – 150 milliseconds latency from read to render (rendering circa 20 million points per second)

For 50km2 tile retrieval and render permitting seamless real-time scrolling over very large AOIs with the ability to then render very large areas at lower spatial resolution (more data into fewer voxels) or conversely stream in the edges and permit different areas to be loaded.

“It’s only when we start tying all these systems together – that we truly get to the amazing capabilities Polaron has to offer – as the inputs into the simulation from the real world permit simulation – rooted in ground truth – and can be used as a ground truth to train Ai and inform the validity of different simulations and models. In effect your able to turn the real world into the training DOJO that ai needs – allowing not just changes but proposals to be modelled and simulated which creates huge opportunities for our Customers & Clients.”

John Tapsfield – Chief Technology Officer – Urban Hawk Ltd

We have also demonstrated that through API gateways we can ensure interoperability and Mobile Access. This has involved both the connection of different simulations in different instances as well as sending and receiving real-time input from a multi-user real-time access enables collaboration to happen in real-time.

We’ve dubbed some of this as being equivalent to being the “Miro of Mapping & Simulation”.

Within 5G – part of the challenge of Public Transport lies in the need to tie into other systems and networks – it’s therefore part of our larger efforts to produce an API portal with a simple and easy-to-use user interface as well as supporting mobile scanning using COTs hardware, as well as larger SLAM and Lidar scanners for more detailed surveys.

Consequently the range of inputs and interoperability has expanded dramatically.

This permits other systems to interface, query and even draw data from Polaron and the simplified experience allows Polaron to engage a larger audience of staff – irrespective of where the data has originated – be it sensors, staff manually adding in situationally aware data and even inviting the public to add and help map the environment.

This has taken us to many different potential applications and push and pull factors; from emissions monitoring and consideration of air quality, through to pulling in thermal data from LEO to build a picture of the environment, consider passenger safety and provide a mechanism for feedback.
All of which is highly relevant to an Insurance application – for both insured and un-insured products.

Ultimately this is the basis for driving preferable customer behaviors and producing the intelligence to alter customer and commercial behaviors to drive better operational practices, plan changes, and evaluate impacts and help to decarbonize the network and meet NetZero targets for both safety and combating climate change.

With the project having been extended due to Covid, this has afforded us more time to extend the amount and scale of data added into the simulation and refine some of the APIs & Data feeds we push into Polaron.

Ultimately this allows a user to add and extract data from Polaron as well as issue instructions and controls – which permits a huge capability to automate and manage your simulations from existing tools, infrastructure, and datasets.

Most recently we are exploring having an external Graph Database and accompanying interfaces whether relational or non-relational databases or even flat data files. The performance we’ve pushed has allowed us to import huge volumes of data at very high speed – then process that within Polaron to generate spatial and relational insights. Given the amount of data available – and sometimes the less than ideal tagging, partials, broken or missing records – a graph lets us not only pull in structured data but relational data – allowing us to infer and provide insight into systems and the cyber-physical relationship to the 3D world they exist in.

Finally WebRTC interfaces Web RTC has been custom integrated into Polaron allowing multiple users to stream and interface with a single or multiple simulations and collaborate via a Miro type experience – WebRTC can also be used as a data or transport layer – allowing data to be added / shared or streamed to other services. Multi-user concurrent sessions We work better together – create a simple means to connect to any contemporary browser.

We have undertaken some tests with Fraunhofer Fokus to establish how we can push our simulation data into their edge based Unity models. Offering a means to engage with these data driven simulations via AR / VR – and offering a fully textured and polygonised experience where appropriate.