Location location location: Positioning’s next act (Part 2)
Every day thousands of users find themselves frustrated by these situations. This problem is not only limited to Lime and Uber, but multiple forms of mobility.
Part 1 of this post outlined the shift toward accurate positioning and the benefits it has for micro mobility, ride sharing, last mile delivery and personal navigation. Part 2 will dive deeper into this theme and explore the key features that a positioning solution requires to enable these use cases. As you’ll see below, it’s no easy feat, but the rewards are well worth the effort.
Three things made GPS the success it is today: low cost, scalability and robustness. In other words, GPS is easily and cheaply integrated into a wide variety of platforms and works globally under challenging conditions. A new solution would need to add value, without sacrificing the qualities above. That value is positioning and orientation accuracy.
Why position and orientation matter
The majority of use cases in urban environments require one meter of accuracy. Meter-level accuracy enables pinpoint location of scooters and bikes, coordinating ride sharing pickup and drop-off points and couriers to deliver items right to your door. However, simple meter-level location accuracy (latitude/longitude/height or position in technical jargon) isn’t enough. What’s truly needed is meter-level location and orientation.
Understanding orientation, meaning the angle and direction of sight from the viewpoint, is what enables technology to fully interact with our surroundings. Knowing what you’re looking at is just as important as knowing where you are.
Position and orientation combined promise to end scooter hunting and awkward conversations about where to meet with your Uber driver. Additionally they enable new use cases such as augmented reality (AR) based navigation for pedestrians as well as drivers. AR based navigation offers, for the first time in 10 years, to transform the navigation UX.
Maintaining ubiquity
Cost and scalability are key for a ubiquitous solution. The higher the cost, the smaller the number of relevant use cases. The cost/benefit of superior positioning must be clear to mobility operators. Adding high cost hardware or HD maps isn’t an option for most (This is a key reason why HD maps uses are currently restricted to autonomous vehicles, which you can read about here).
Scalability depends on a flexible solution that can work across different hardware, operating environments and geographies. In this case, a software first solution is the only kind that can scale. Additionally, any solution must work regardless of changing weather, connectivity, lighting and geography.
The power of sight
Providing a scalable superior positioning technology will require addressing the attributes described above. Various approaches developed to date fall short. This includes HD maps (cost), simultaneous localization and mapping, i.e SLAM (cost and robustness), and even long-awaited enhancements to the current GPS such as the EU’s Galileo project (no orientation information).
A solution to solving this problem seems to lie in mimicking how we solve these situations every day — eyesight. Vision is the most powerful of our senses, and therefore implementing it on a technological level offers an intuitive approach to tackling these problems. Recent technological advancements in computer vision and AI are finally enabling a viable humanized Visual Positioning System (VPS).
Enter VPS
Visual Positioning Systems aren’t new. Over the past few years various attempts have been made to implement this technology but limitations in AI, edge processing and computer vision prevented them from becoming truly viable. Today, most of these challenges are surmountable. In recent years, major advances have been made in the areas of AI and computer vision, and the processing power of many edge devices has increased to the extent that a computationally light solution can be implemented.
So what would a scalable VPS solution look like?
1. Accuracy: Achieving ~1-meter accuracy for positioning and orientation.
2. Low cost: A scalable VPS would need to run on low cost hardware. This means the algorithms must be computationally efficient and light, no GPUs.
3. Scalability: To attain scale, the solution must be software focused, built to run on commodity hardware found on edge devices such as vehicles and smartphones.
4. Robustness: Any VPS must be resistant to changes in the environment, including: different weather patterns, construction, and obstructed field of view. Additionally, constant connectivity to the cloud can’t always be relied upon.
This list of requirements makes it evident why such a solution hasn’t yet become viable. Creating a software-only solution that works despite such challenging and changing environments while running on commodity hardware is no easy feat.
Despite these challenges, such a solution could indeed be the technology powering modern mobility, making it a challenge worth tackling.
If you’d like to find out more about our scalable VPS technology, feel free to reach out.
Yehoshua is the Director of Business Development at Spatial Logic, an Israeli startup developing a Visual Positioning System for the mobility ecosystem. Before Spatial Logic, Yehoshua led Business Development at OurCrowd, a global venture capital firm, where he worked closely with Honda, Shell and Denso on the mobility sector.
