Last mile delivery (LMD) is a critical segment of the logistics industry. The size of LMD was valued at over 30 billion dollars in 2018, and will grow significantly through 2025, where it is projected to be worth over 60 billion dollars. However, LMD faces severe inefficiencies, costing firms a significant amount of revenue annually.
Today, technology companies such as Doordash and Postmates are entering the last mile delivery sector – they bring unprecedented convenience to everyday consumers. Nevertheless, LMD is costly – these businesses operate on a loss consistently. There needs to be a solution to allow these firms to be profitable to benefit society in the long run.
Delving into the mechanisms behind this inefficiency – human drivers place a significant burden on operating expenses. Breaking down the cost of parcel distribution– each delivery costs about $1.60 per ride. However, through replacing humans with robotic-driven vehicles, this number is reduced significantly to just six cents per ride.
Attaching this to a real life use case – Doordash completes ~300 million deliveries annually. At the same time, the company lost $450 million dollars in 2019. However, with the cost benefits of autonomous vehicles (AV) outlined above, the company can save over $460 million dollars with AVs replacing human drivers, allowing it to experience growth for years to come.
Given the billions of last mile deliveries completed annually, for technology companies involved in LMD to be profitable in the long run, they must adopt a delivery model that involves robots rather than humans acting as drivers. However, solutions today can only enable partial autonomy – they are based on utilizing legacy technology, such as the GPU. Because these platforms are not purpose-built for the monumental task of self-driving, a novel, purpose-built solution must be brought to the market to enable fully autonomous vehicles (AV).
When we drive, our brains use a data-center level of compute while consuming minimal power to allow us to quickly and accurately perceive our surrounding environment in real time. For an autonomous vehicle to mimic this, it must be equipped with a self-driving platform that can compute at least 75 Tera-Operations-Per-Second (TOPS) for every watt of power consumption. This immense efficiency requirement, known as the visual perception problem, is the largest barrier to realizing full autonomy.
We @ Recogni are developing such a product. Through leveraging key innovations in math, ASIC architecture, and artificial intelligence, our solution brings unmatched compute and efficiency to the market – 1000 TOPS of processing power while consuming less energy than a lightbulb. This best-in-breed platform is purpose-built to solve the visual perception problem outlined above and enable fully autonomous vehicles. Through this, companies in the last-mile-delivery sector can replace their current human-driven model with robotic delivery, allowing them to be profitable in future.
By Sidhart Krishnamurthi, Product Management @ Recogni
Realtime Object Recognition
Original. Reposted with permission.