Bhopal Entrepreneur Builds L-5 Autonomous Driving System In India

In the realm of technological innovation, few advancements promise to revolutionise daily life as profoundly as autonomous driving. While self-driving cars have become a reality in several countries, India's readiness for such groundbreaking technology remains a subject of debate. 

Some argue that autonomous vehicles will have an adverse impact on employment and labour, while others feel that roads in India are not yet ready for autonomous vehicles. Nitin Gadkari, Union Minister of Road Transport and Highways, has reportedly said, "Driverless cars should never be allowed in India as they would cost the jobs of several drivers." 

According to a report by Fortune Business, the global autonomous cars market was valued at $1.45 billion in 2020, which touched $1.64 billion in 2021. The number is expected to reach at least $1 Trillion by 2028. 

The Indian autonomous vehicle market is projected to grow at a rapid pace. According to the founder of Swaayatt Robots, Defence sector alone is going to be $50 billion by 2030. These vehicles are expected to face challenges like chaotic traffic conditions, inconsistent road infrastructure, and diverse weather conditions.

Tesla is one of the names that crosses our minds when we first hear the term' autonomous vehicles'. India is on a mission of technological advancement, and many startups are ready with the prototypes of self-driving cars. 

Swaayatt Robots is one such startup that took to X (formerly known as Twitter), a video featuring a self-driving four wheeler navigating the challenging traffic conditions in Bhopal. The video shows how the autonomous vehicle was maintaining a safe speed. 

The vehicle detected obstacles and was easily turning around from them. The car had a driver for safety as it was a test vehicle. The autonomous driving company has claimed that it achieved the world's first Level 5 autonomous driving capability. This means that the vehicle can drive on a road with traffic and under all conditions without being driven by human beings. 

Sanjeev Sharma: the Man Behind Swaayatt Robots

Swaayatt Robots founder Sanjeev Sharma

Sanjeev Sharma, Founder & CEO of Swaayatt Robots, a visionary entrepreneur with a passion for robotics and artificial intelligence, embarked on the journey of autonomous navigation during his fourth semester of engineering in 2009. 

Sharma was pursuing B. Tech in Electrical Engineering at IIT Roorkee. He completed his master's in computer science from the University of Alberta in 2014. 

In a conversation with Startup Pedia, Sanjeev Sharma recalled, "I came back to India deferring the Computer Science PhD offer at the University of Massachusetts Amherst in the US to 2015. I wanted to research further and introduce autonomous driving in the world's most complex traffic and environmental conditions imaginable, which happens to be my  home country, India." 

Sharma was inspired by DARPA (the R&D agency of the US Department of Defense) autonomous driving challenge videos and how  Pieter Abbeel under Andrew NG at Stanford University built autonomous flying machines using reinforcement learning. As his curiosity made him look more into this challenge, he stumbled upon several videos on YouTube by MIT on autonomous driving technology for the DARPA Urban Challenge.

"I was determined to solve the problem of autonomous driving in the most challenging of environments and traffic scenarios for autonomous vehicles. I came back and decided it was time to start the company, and in May 2015, I registered the company," added the Bhopal tech entrepreneur. Sanjeev has been interested in robotics since 10th grade.

Sanjeev Sharma has devoted nine years of his life into cutting-edge research to develop autonomous driving technology. He started self-study during his undergrad days and read several advanced books and numerous research papers in machine learning, mathematical optimization, robotics, and reinforcement learning. His core research was on motion planning and artificial intelligence. The tech entrepreneur took up projects to create motion planners that help navigate Indian traffic.

The founder of Swaayatt Robots founded another startup named Deep Eigen in May 2020 and formally registered it in 2021. It is an online education platform that offers graduate-level engineering, science, and mathematics courses. He is also a recipient of the Leading 40 Under 40 Data Scientists in India award at the Machine Learning Developers Summit for his research in autonomous driving technology.

Swaayatt Robots:

The Bhopal-based tech startup focuses on  solving the Level-5 autonomy problem, with an arduous objective of solving the Level-4 autonomy problem by the end of the year 2024. The aim is to make travel safe, robust, cost-effective, and mapless. It offers AI software platforms and solutions, autonomous driving and industrial and warehouse robotics. 

The Bhopal-based startup has developed, as claimed, level-5 autonomous driving technology that can operate entirely independently. Additionally, it wants to conduct a compelling 100 KM/H autonomous driving demonstration on Indian roads. Swaayatt Robots is heavily involved in reinforcement learning and unsupervised deep learning methodologies. 

An autonomous vehicle needs to process the data from visual sensors, like cameras, RADARs and LiDARs to understand its environment and build a 3D model of the world around it. The AI software to enable this is known as the perception software in this field. 

Swaayatt Robots, over the years, have developed computationally efficient AI perception software that allows autonomous vehicles to visualise its environment using only off-the-shelf cameras, the tech startup claims. The company says that for off-roads autonomous driving, for military application, they may end up using LiDARs, as can be seen in their demos off-roads, but in August they will showcase a camera-only technology for off-roads as well, the founder claimed. 

Typically autonomous vehicles require high-definition maps of the environments to function. Swaayatt Robots develops its own HD maps for last-mile autonomy applications. They have developed the technology that in general doesn’t require high-definition maps of the environments, and have been a first mover in this regard.

Localisation algorithms utilise visual and other sensory data to determine a vehicle's position relative to either a GPS-only map or a detailed, high-fidelity map, depending on the demos and the capabilities they want to highlight. Planning algorithms then compute the motion and behaviour of the vehicle to be executed by the controllers.

Motion planning and decision making is the most heavily focused research area at Swaayatt, and it accounts for 70% of their R&D, the founder said. Their focus has been on developing sophisticated AI based decision making systems capable of negotiating chaotic traffic scenarios, as can be seen throughout their demos.

The technology is capable of negotiating bi-directional traffic on a single lane, a capability unique to Swaayatt Robots. Their current primary research focus lies in developing a decision-making system that possesses the context of the world, eliminating the necessity for explicit computation to identify diverse obstacles, their intentions, or road delimiters. 

Explaining the technology, Sanjeev Sharma said, "Autonomous driving is a combination of various different branches of engineering, science, and mathematics. We ensure the operational cost of the overall system is lower compared to existing competing technologies from prominent competitors." 

The tech startup founder started with an investment of Rs 80 lakh. He further added that the biggest challenge was the cost of setting up the company.

The autonomous driving research startup had raised a total funding of $3 million in a single round in 2021- from an investor in the US. Followed by the most recent angel round on July 16, 2021, securing $3 million from investors based in the United States. This investment was made at a valuation of $75M. 

USP of Swaayatt Robots

In off-road autonomous driving, one of the startup's current research frontiers is exploring embodied intelligence, where the system learns directly from human demonstrations. However, this differs from behaviour cloning. For instance, startups like - Comma AI are engaged in behaviour cloning.

In behaviour cloning, the machine is exposed to every possible scenario of success and failure while being driven, with the aim of learning from these scenarios. However, recovery becomes impossible if the machine fails or goes beyond its limits in behaviour cloning.

To address this, reinforcement is necessary. However, integrating reinforcement poses significant challenges. Expertise in the field is essential due to its complexity and the high level of competition.

The tech startup founders said that Tesla is one company pioneering autoregressive RL. In this approach, data is collected from hundreds of thousands of drivers operating the vehicles. Instead of instructing the machine to mimic behaviours, a reinforcement signal guides the learning process, highlighting the significance of autoregression.

However, only companies with Tesla's level of resources and expertise can effectively implement autoregressive RL. It's a formidable and highly competitive approach.

"Since the inception of our company, with $3 million in funding, I, as the researcher, have specialised in reinforcement learning. We have initiated research into data-efficient RL," said the founder of Swaayatt Robots. 

In essence, the tech startup's pioneering work revolves around developing data-efficient models in RL, regardless of the learning methods employed. Whether it's inverse learning, forward learning, or apprenticeship, the ultimate goal remains the same: creating efficient reinforcement learning models. This is a challenging task, as evidenced by the scarcity of such models globally.

"Our recent research efforts have been focused on exploring avenues to automate processes like map generation, particularly in the mapping industry. While companies like TomTom, DeepMap, and Google 3D rely on human labelling for map building, we're exploring theoretical frontiers like transfer learning to streamline the process," added the Bhopal-based tech entrepreneur.  

Future of Swaayatt Robots

Previously, Swaayatt Robots initiated autonomous driving trials on city roads in India during the nighttime. Now, recently, it has tested the technology in broad daylight. The tech revolutionary mentioned that his team is working on an undisclosed 'X' paradigm, intending to showcase how autonomous vehicles will independently acquire various skills. 

Furthermore, according to Sanjeev Sharma, "our aim is to capture 25% of the Global market share by 2030 and transition into a multinational corporation." 

He is confident that Swaayatt Robots will establish a widespread presence across India within a year. The founder of the Bhopal-based tech startup believes that in the next year, they will raise funding worth $ 1 billion.  

In conclusion, the rise of Swaayatt Robots depicts India's growing prowess in the field of autonomous driving.

FAQs 

What is Swaayatt Robots, and what does it specialise in?

Swaayatt Robots is a Bhopal-based tech startup specialising in making connected autonomous driving technology accessible, affordable, and available globally. It offers AI software platforms, solutions for autonomous driving, and industrial and warehouse robotics.

Who is the founder of Swaayatt Robots?

Bhopal-based tech entrepreneur Sanjeev Sharma is the founder of Swaayatt Robots. 

What is Level 5 autonomous driving, and how does Swaayatt Robots claim to achieve it?

Level 5 autonomous driving refers to a vehicle's ability to operate entirely independently on any road condition without human intervention. Swaayatt Robots claims to have developed level 5 autonomous driving technology, which they demonstrate through vehicles that navigate challenging traffic conditions autonomously.

What are Swaayatt Robots's future plans and goals?

Swaayatt Robots aims to capture 25% of the Global Market share by 2030 and transition into a multinational corporation. They plan to establish a widespread presence across India within a year and aim to raise funds worth $1 billion in the next year. Their ongoing research focuses on showcasing how autonomous vehicles can independently acquire various skills.