Avideo of a Maruti Alto K10, driving by itself in what seems to be a busy highway went viral among Indian social media users. In the video taken from inside the car, users can see the mighty little Alto K10, steering itself, with no one touching the steering wheel. In its truest sense, this was, a self-driving vehicle
It turns out, this enormous engineering feat of giving the humble Alto self-driving capabilities, by a Bengaluru-based techie, named Mankaran Singh.
In an interview with Analytics Magazine India, Singh explained how he went about modifying the car and refining the self-driving system. He transformed his modified Maruti Alto K10 into an autonomous vehicle using a second-hand Redmi Note 9 Pro and Flowpilot, an open-source driver assistance system.
Flowpilot is a derivative of Comma.ai’s OpenPilot and is compatible with Windows/Linux and Android-powered devices. Singh used learning models from Comma.ai, as training driving models independently would require extensive data and significant financial resources.
The inspiration for this initiative came from George Hotz, the founder of Comma.ai, known for enabling autonomous vehicles using smartphones and proprietary devices. Singh, an avid programmer and autonomous systems enthusiast, is concurrently engaged in gaming, developing web applications, and working on Flow Drive, his passion project behind Flowpilot, while also being part of Ola.
Flowpilot functions as Adaptive Cruise Control (ACC), Automated Lane Centering (ALC), Forward Collision Warning (FCW), Lane Departure Warning (LDW), and Driver Monitoring (DM) for a growing list of supported car makes and models. It records data from various sensors and logs, providing compatibility with a wide range of smartphones that support OpenCL.
Singh emphasized that the performance of Flowpilot depends on the phone’s quality, stating that a mid-range phone is sufficient for reasonable performance.
However, Flowpilot supports over 200 car models, including popular brands, but currently does not support cars in India, natively, meaning, you will have to get a little creative with it to get it to work.
Despite this, users in India prefer Flowpilot due to its superior performance compared to stock car systems.
So how does this differ from Tesla? Well, for starters, Tesla’s long-term vision involves complex neural networks, and their EVs, usually use eight cameras, which results in a substantial computational load, because of which, it can be operated within city limits and on extremely busy roads. Flowpilot on the other hand, focuses on highway driving automation using a minimal camera setup and additional sensors.
The immediate application of Flowpilot involves highway driving automation, aligning with the growing prevalence of Advanced Driver Assistance Systems (ADAS).
Singh emphasised safety features, such as automatic emergency braking and blind spot monitoring, showcasing the system’s ability to operate at a higher rate than human processing.
However, Flowpilot has limitations, particularly in the context of Indian roads. Factors such as low-light conditions, shadows, and unpredictable obstacles pose challenges.
As technology advances, we will see more and more autonomous vehicles being deployed in controlled environments like college campuses and tech parks. Until then, we will mainly see autopilot systems deployed on large fleets that mainly operate outside city limits.
(With inputs from agencies)
