Swaayatt Robots, a Bhopal-based autonomous driving research startup, raised $4M on June 3rd at a valuation of $151M. This fund is part of their larger second Seed round of $15M. The startup aims to obtain the remaining $11M at a valuation of around $175M-$200M soon; investors in North America, Europe, and Australia have expressed interest.

Founded by Sanjeev Sharma, Swaayatt Robots previously raised $3M from a US-based investor in 2021 at a valuation of $75M. Although the startup has not revealed the investor’s name, Sanjeev confirmed the investor’s participation in the current round as well.

In the next 6 to 7 months, Swaayatt Robots plans to raise $50M in pre-Series A to expand its global footprint and scale up the technology significantly. The startup is initially targeting operations in North America, the UK, and the Middle East.

“We want to solve the Level-4 autonomous driving problem globally at scale, fueled by our Level-5 AI models and algorithmic frameworks for autonomous driving,” says Sanjeev to Analytics Drift. Going forward, the startup is heavily going to invest in “(i) doing cutting-edge R&D in unsupervised learning and reinforcement learning domains to robustify the perception and planning capabilities, and (ii) bridging all the AI models and algorithmic frameworks the startup has developed, to make an architecture that can be scaled globally for Level-4 autonomous driving”, the founder highlighted when speaking to Analytics Drift. For such ambitious targets, the startup is also planning to raise $1.5B, beyond pre-series A, in the next 15 months. Sanjeev also believes that Swaayatt Robots is poised to solve the Level-5 problem and emerge as one of the major technology suppliers for autonomous navigation worldwide by 2028.

With the recent funds, Swaayatt Robots will invest in R&D to further enhance the development of autonomous vehicles for both on-road and off-road conditions. One of the pioneers in LiDAR-less navigation, the startup has showcased several demos of vehicles effortlessly navigating uncertain terrains.

For instance, in one of the demos, the startup exhibited the ability to negotiate the incoming flow of traffic off-roads, a technological capability that is currently unique to Swaayatt Robots at this time. Even companies like Kodiak Robotics, Overland AI, and the US DARPA Racer program’s participants have struggled to showcase similar capabilities. This has been a result of years of cutting-edge R&D in deep learning, reinforcement learning, motion planning and decision making, machine learning, and other frontiers of theoretical computer science and applied mathematics.

Over the years, Swaayatt has strived to be the torchbearer for solving complex problems in the autonomous vehicles industry. Even in late 2023, they displayed several ground-breaking innovations. Last October, the startup demonstrated bidirectional traffic negotiation on single-lane roads—a capability again unique to Swaayatt Robots. 

Backed by impactful R&D and successful demonstrations, Sanjeev and his team aspire to solve the Level-4 problem globally. “In India, we have already demonstrated several Level-5 capability algorithmic frameworks to solve certain frontiers of problems in autonomous driving. For example, in our March 2024 demo at the Baglamukhi Mata Mandir, we also demonstrated the ability to cross unmanaged intersections. Typically, even crossing a managed-traffic-light intersection is considered a challenge in the US by major companies like Waymo, Cruise, Tesla, etc.,” asserts Sanjeev.

Swaayatt will continue to exhibit several demos of its autonomous driving capabilities in the coming months. For example, in August this year, the startup plans to highlight major capabilities previously unseen in the field of autonomous driving at large. For last-mile autonomy applications, Swaayatt has been conducting R&D to develop sparse maps and inference algorithms that have very low computational requirements, along with automating the generation of high-definition feature layers in the maps.

Sanjeev thinks that one of the core challenging problems in the autonomous driving industry is safety and operational cost. While the startup will continue to invest in enhancing the models to ensure safety in the presence of traffic-dynamics that is highly stochastic, complex, and adversarial in nature, Swaayatt has developed efficient models to reduce operational costs. 

“Over the years, we have been demonstrating our deep learning algorithmic frameworks, in the areas of perception and planning, that are an order of magnitude computationally efficient compared to the state-of-the-art while having better performance and more capabilities. Going forward, we will unify most of the algorithmic frameworks we have developed into holistic autonomous agents that are 20-30 times computationally efficient in holistic decision-making and planning for autonomous vehicles. 

For example, the current version of our motion planning and decision-making algorithmic framework, which we have been demonstrating off-roads, runs at more than 200 Hz on a single thread of a laptop processor. We are further extending this with the integration of deep reinforcement learning. It will eliminate the need for explicit perception algorithms required for off-road navigation and will operate at close to 40 Hz. This is just one of the instances of the several frameworks we have been demonstrating over the past few months,” explains Sanjeev.

Sanjeev also believes that we need a solution to autonomous driving in the presence of highly stochastic, complex, and adversarial traffic-dynamics on the roads to come up with Level-4 or Level-5 autonomous driving technology. Without this, safety cannot be ensured and will require endless iterations and discovery of corner cases. Therefore, Swaayatt Robots is solving the hardest AI problem of this decade, enabling autonomous agents to learn and negotiate adversarial, complex, and stochastic traffic-dynamics. 

By solving the root cause, the idea is to eventually get numerous by-products and make the technology ready for several verticals, such as (i) warehouse autonomous navigation technology, (ii) campus autonomous navigation technology, and (iii) autonomous trucking on highways. Sanjeev, while speaking to Analytics Drift, mentioned that “the core focus, however, of the startup is going to be doing cutting-edge R&D in various frontiers of modern AI, theoretical computer science and applied mathematics, to develop foundational models to solve the problem of autonomous general navigation, that enables autonomous vehicles to safely navigate from point to point while being operationally cost-efficient.”

With such competencies, Swaayatt Robots is now working with major OEMs to commercialize the technology later this year.