SASKEN is pioneering R&D and innovation in various sectors like industrials, automotive & consumer electronics. Technologies like AI form the core of its value offerings. Amit Verma, Senior Solutions Architect - AI Portfolio at Sasken spent some time elaborating on how the journey has been for SASKEN transitioning to machine led intelligence from chip-based intel. “With a long-standing industry expertise in chipset and communication technologies, the leap to developing cognitive solutions was a natural one. The rapid pace at which silicon technologies have evolved, including open source software platforms such as Android, has made it possible to leverage computing and low-latency high-bandwidth connectivity to make a wide range of cognitive solutions available today,” says Verma.

Adept at ADAS

The demand and development of high-end cognitive solutions has been fuelled by rapidly advancing computing abilities, digital technologies and sensor networks. With its established track record in engineering, Sasken has navigated successfully into domains like automotive, industrials and consumer electronics - in automotive, the company has devised solutions such as Driver Drowsiness Detection, Traffic Sign Recognition, and Forward Collision Warning using image processing and Deep Learning Neural Networks for decision making. They have also made headway in cognitive machine learning framework integration and development of a reference solution for Voice-Enabled Security Commands, Voice Recognition, and NLP.

ADAS is a particularly heightened focus area for Sasken, with their solutions being well received by customers in North America, Europe, and Japan. Recently, a Japanese Automotive Tier-1 provided Sasken the opportunity to create an Adaptive Driving Beam system integrating Digital Light Projection (DLP) that reduces glare on pedestrians and cyclists and ensures more precise control of light distribution through independent LED control. Sasken worked on the porting, optimizing, and developing pedestrian detection algorithm using Deep Learning models.

Overhauling consumer electronics and heavy industries with ML and DLNNs

In heavy industries space, Sasken has helped develop an industry-grade machine vision camera with advanced real-time image processing and analytics on the edge. The camera is equipped with a higher percentage of factory automation and automated real-time checks for defect identification and other quality KPIs. Meanwhile, in consumer electronics, the company equips appliances with vision and voice-based interactions enabling them to recognize the users and content being processed. Other PoCs that Sasken’s engineers are working on include Remaining Useful Life (RUL) to predict lifespans and provide early warnings of failure of industrial equipment with the purpose of reducing downtime in the industrials segment, using AI to differentiate anomalous sound from normal sound to indicate damage to a machine and a CV-based multi-camera surveillance system.

A rich IoT partnership for augmented AI capabilities

A comprehensive, end to end suite of AI solutions can thrive with an ideal product and hardware partnership. That’s what MediaTek IoT is for Sasken. MediaTek’s Rich IoT program is spurring AI innovation by enabling enterprises, startups, emerging brands and leading OEMs access to a wide range of IoT devices with edge-AI capabilities, and helping them address untapped market opportunities. “We are leveraging our extensive knowledge of silicon platforms, connectivity, imaging, and smartphone integration to make home and industrial appliances more intelligent.” At CES 2020, Sasken demonstrated an end-to-end prototype of a smart refrigerator with a smart panel, powered by a MediaTek chipset. This panel allows users to know what items are located inside the refrigerator, their quantities, and provide serving suggestions without having to open the refrigerator.

Finding, retaining and upskilling talent in an AI-first world

The new world order is a concept that Verma is too aware of. Despite having decades of expertise under his belt, Verma still thinks there’s plenty to learn. “It is necessary to keep abreast of the latest technology trends because hardware and software, related to AI, can get obsolete very fast.” he says.

And this doesn’t end with senior leaders like him. He leads a team of 20+ engineers, and admits that finding and grooming the right talent for a niche area like AI today is not easy. Sasken invests a substantial amount of time and resources building inhouse subject matter expertise. “The ability to offer them problems that find applications in the real world is key to getting them integrated into the core delivery engine that understands and solves customer problems.”