In the past five years alone, major automakers and technology giants have invested $50 billion in developing autonomous vehicles. This race to achieve vehicle autonomy and to be among the first to launch commercially available autonomous vehicles has resulted in rapid innovations in chip design, sensors, AI-based computer vision-based solutions, etc., which are crucial for realizing vehicle autonomy.

Artificial Intelligence and Cognitive Sciences are powering latest ADAS/Autonomous solutions fusing and interpreting data from multiple sensors. However, this presents various challenges to Automotive manufacturers such as managing the complexity of neural networks, cost of developing and validating ADAS solutions, lack of control on core technology, complexity in porting solution to semiconductor platform integrating cameras/Radar/LiDAR, improving test coverage, etc. Thus, they are challenged to respond more quickly to a larger number of applications and shorten development cycles, while retaining quality and keeping their overall investment under control.

Leveraging its partnerships with leading silicon vendors, Sasken enables OEMs and Tier-1s to accelerate product development by using Perception Engine/Module that integrates computer vision, radar, lidar sensor, connectivity, mapping and path planning technologies.

Trends Implications Challenges Sasken is Solving

Regulations mandating compulsory ADAS features on all new vehicles

There is a need to deploy ADAS even for vehicles targeted towards lower market segment

  • Ensuring minimal hardware changes to provide additional ADAS features and minimizing validation costs
  • Minimizing complexity and pruning neural nets for edge use cases

Emergence of Adaptive AUTOSAR as a choice for ADAS/V2X domain

Need to develop safe, secure & time critical ADAS/V2X ECUs based on Adaptive AUTOSAR architecture

Designing safe and secure Adaptive AUTOSAR based ECU architecture

New safety and security standards for Autonomous vehicles

Need to adhere to safety and security standards like ISO 26262, ISO 21448, ISO 21434, UL 4600

Meeting both safety and security goals and compliance

Increased focus on performance and accuracy to ensure ZERO fatalities

There is a need to develop efficient and faster DL-based algorithms on high performance computing platforms (DCUs)

  • Reproducing data from various sources for improving accuracy of algorithms
  • Selecting right set of parameters for measuring performance

Increased customer apprehension towards Autonomous vehicles

Vehicles need to be made drive ready in any environment, i.e. terrain, weather, traffic scenario

Improving test coverage to cover the edge-scenarios as well

Emergence of 5G Connectivity technology

  • Provides an opportunity to off load non-streaming use-cases and compute to cloud.
  • Creation of data management platform to accelerate algorithm testing and training

Analyzing vast amounts of on-road test data


  • OnBoard Intelligence Applications:
    • Driver Alertness Detection, Forward Collision Warning
    • Traffic Sign Detection & Road Sign Detection
    • Range Sensing, Pedestrian Warning, Object Detection
    • Lane Detection/Drift, Blind Spot
    • Surround View/Point Cloud
    • Parking Assist
  • Connected Intelligence Applications:
    • V2V Coordination, V2I Interaction
    • HD Maps, SLAM, Collision Avoidance
    • Lane Change Assist & Route Planning
    • Micro Path Planning & Traffic Management
  • Automated Vehicle Control (Low Speed) Applications:
    • Automated Parking
    • Automated Emergency Breaking
    • Automated Steering Control
  • Engineering Services to develop platforms for ADAS/Autonomy with more control on core technology:
    • Heterogeneous Connectivity & Computing
    • Sensor & System Integration
    • Point Cloud Processing & Localization
    • System Validation and Maintenance
  • Reference Hardware and SDK design & development services
  • Data management platform for both on-cloud deep learning and validation services
    • Remote vehicle update services (OTA)
  • Safety and Security
    • Verification & Validation activities as per ISO 26262, ISO 21448 (SOTIF), ISO 21434, UL 4600

System Architecture of Sasken’s ADAS Platform


Click to Enlarge


  • Experience in developing and productizing Vision and RADAR fusion perception system integrating on-board intelligence applications that is:
    • Configurable and scalable to accelerate ADAS product roadmap to L3 autonomy
    • Provides easy personalization with smart fusion of multiple sensors with Deep Learning on Automotive Grade Platform
    • Unified solution that lowers the cost of ownership and reduces add-ons for partners
    • Market-ready having trained with more than 10,000 hours of test drive data in India & Japan
  • 40+ person years of sensor and computer vision expertise
  • Experience in turn-key product development with state-of-the-art labs with Deep Learning training, Autonomous Vehicle Sensors, and Test vehicle
  • Customizable and reusable framework and solutions



Testing of automotive components i.e., its verification and validation...

The goal of AUTOSAR is to not only introduce standardization in the de...

Growth of Data Science in Vehicles In recent times, the need for comfo...

Hardware and Software architecture in vehicles in recent decades Since...