Pioneering AI Use Cases for the Automotive Industry

Product

AI-based Advanced Simulations

Product Development

Tata Elxsi Use Case

Use Case: ML-based Transfer Function Prediction for Electric Drive Units (EDUs)

How it Works: Using Machine Learning (ML) to predict the transfer function for Electric Drive Units (EDUs). This involves training a model on performance data to learn the relationship between inputs and outputs.

It enables accurate predictions for new inputs, ensuring a robust design at early development stages. Transfer functions for specific or multiple paths can be set, reducing manual prediction time from 40 hours to 5 minutes.

Tech Stack

Data Collection Tools

  • Excel

Analysis Software

  • MATLAB and Python

Simulation Tools

  • Finite Element Analysis (FEA) software (e.g., HyperMesh, Optistruct)

Test Targets

  • Acceleration V/s Frequency

Testing Equipment

  • Anechoic Chambers
  • Vibration Test Rigs
  • Sound Level Meters 

Smart Subscription Services for Automotive

Aftermarket

Tata Elxsi Use Case

Use Case: Personalised Subscription Recommendation for Vehicles

How it Works:

The system collects data from driving behaviour, vehicle usage, and user preferences. Machine Learning algorithms analyse this data to identify patterns and predict the most suitable subscription services for each user. ​

​The recommendations are dynamically updated based on real-time data and user feedback, ensuring they remain relevant and personalised. By delivering customised suggestions, the system helps users choose the most appropriate and cost-effective services for their specific needs.

Tech Stack

  • Machine Learning Models ​
  • Data Processing and Analysis: Python libraries like Pandas or NumPy ​
  • Real-time Data Handling: Apache Kafka or MQTT​
  • Cloud Computing: AWS​
  • APIs and Integration: RESTful APIs ​
  • Recommendation Engine

LEXI Connected Digital Twin Platform

Aftermarket

Tata Elxsi Use Case

Use Case: GenAI-based Chatbot for Digital Twin

How it Works:

The GenAI-powered chatbot interfaces with digital twin systems, leveraging natural language processing (NLP) to understand and respond to user queries. ​

​Users can ask questions about the digital twin’s current state, performance metrics, predicted maintenance needs, and potential issues. The chatbot processes these queries using advanced NLP models and retrieves relevant information from the digital twin’s data repository. The generated responses make complex data easily understandable and actionable for users. ​

Tech Stack

  • Natural Language Processing (NLP) ​
  • Machine Learning Models ​
  • Private LLM​
  • Cloud Computing: AWS​
  • Chatbot Frameworks​
  • APIs and Integration: RESTful APIs ​
  • Digital Twin Framework​

SignaTE

Aftermarket

Tata Elxsi Use Case

Use Case: AI-Driven Predictive Maintenance for Three-Phase Motors​

How it Works:

Tata Elxsi’s SignaTE platform uses sensors on three-phase motors that collect real-time current, monitoring for anomalies that indicate faults and incipient faults. ​

​Machine Learning and Deep Learning techniques analyse this data to identify common issues like bearing wear, rotor bar defects, and insulation degradation. AI models trained on historical and real-time data predict motor faults with high accuracy, continuously improving through self-learning mechanisms.

Tech Stack

  • Machine Learning and Deep Learning ​
  • Data Processing and Analysis​
  • Real-Time Data Handling Tools​
  • Secure Cloud Computing Platforms​
  • IoT Platforms​
  • Data Visualisation: Tableau, Power BI​
  • Model Training and Validation​

Experience

Connected Car/Cloud  

Aftermarket

Tata Elxsi Use Case

Use Case: Road Friction Predictor​

How it Works: The module collects data from optical sensors to analyse road conditions, CAN data to understand vehicle dynamics, and cloud-based information for additional context such as weather conditions.

AI/ML algorithms process this data to predict road friction levels. The predicted friction is then communicated to ADAS features, allowing them to adjust vehicle control parameters in real-time, such as braking force and traction control, to enhance safety and handling.

Tech Stack

Data Collection and Integration: 

  • Optical Sensors: Cameras, Lidar 
  • CAN Bus Interfaces
  • Cloud-based Data Sources

AI/ML Algorithms: ​

  • Machine Learning Frameworks: TensorFlow, PyTorch
  • Predictive Modelling Techniques 

Data Processing and Storage: ​

  • Edge Computing Devices
  • Cloud Computing Platform

Integration with ADAS: ​

  • Vehicle Control Systems

ADAS/AS

Aftermarket

Tata Elxsi Use Case

Use Case: Real-time Object Detection and Tracking​

How it Works: ML models are trained on vast datasets of labelled images and videos to recognise various objects and their movements. Real-time data from vehicle-mounted Cameras, Lidar, and Radar sensors are processed by these models to detect and classify objects in the vehicle’s environment.

The system continuously tracks these objects, estimating their trajectories and predicting their intent to avoid collisions and navigate safely. The processed information is fed into the vehicle’s control system, enabling real-time adjustments in speed, steering, and braking.

Tech Stack

Data Collection and Integration: 

  • Sensors: Cameras, Lidar, Radar 
  • Data Ingestion Tools  
  • Ground truthing tools

Machine Learning and AI Algorithms: 

  • Deep Learning Frameworks: TensorFlow, PyTorch
  • Object Detection Models: YOLO, Faster R-CNN 
  • Tracking algorithms: SORT, Deep SORT

Data Processing and Storage: ​

  • Real-time Processing Frameworks: Apache Kafka, Apache Flink
  • High-performance Computing: NVIDIA GPUs, TPUs

Integration with Vehicle Systems: ​

  • Vehicle Control Systems: ROS, Apollo 
  • Communication protocols

Visualisation and Monitoring: ​

  • Dashboard tools: Grafana, Custom Interfaces
  • Simulation environments: CARLA, LGSVL 

In Car-Entertainment​

Research

Tata Elxsi Use Case

Use Case: AI-backed HMI Design for Cluster, Infotainment, and Rear Seat Entertainment

How it Works: This approach leverages AI to enhance the design of HMIs for vehicle clusters, infotainment systems, and rear seat entertainment. AI tools assist with creating research questionnaires, predictive modelling, persona creation, sentiment analysis, wireframing, layout optimisation, and generating graphic components with code.

AI also aids in rapid prototyping, product localisation, colour palette generation, alternate design options, and design recommendations, as well as creating 2D and 3D assets and performing UX/UI testing.

Tech Stack

Research and Data Analysis ​

  • TensorFlow, PyTorch 

​Design and Prototyping​

  • Figma, Sketch 
  • Adobe Sensei 

​Graphic Generation and Localization:​

  • Smartling, Transifex 

​Testing and Validation: ​

  • User Testing, Maze 

​Integration and Automation: ​

  • REST APIs, GraphQL 

Connected Cars

Product Development

Tata Elxsi Use Case

Use Case: AI-Based Driving Score Calculation for EV Range Estimation

How it Works: AI evaluates driving performance by analysing behaviours such as idling, harsh acceleration, braking, and speeding. It calculates a driving score, offering insights to improve driving patterns for better range and fuel economy.

​The score also highlights specific areas needing improvement and compares performance across different vehicles and platforms. Additionally, it aids in vehicle design enhancements and supports a driver coaching system that fosters safe driving habits.

Tech Stack

  • AI and Machine Learning​
  • Telematics and IoT Sensors​
  • Data Analytics Platforms ​
  • Cloud Computing​
  • Mobile and Web Applications ​

Maintenance

Predictive and Prescriptive Maintenance with AI

Aftermarket

Tata Elxsi Use Case

Use Case: Predictive and Prescriptive Maintenance​

How it Works:

The system utilises analytics algorithms to assess the remaining useful life of vehicle components, enabling proactive maintenance scheduling. ​

​Anomaly detection algorithms continuously monitor vehicle data for deviations from normal behaviour, forecasting potential issues. Prescriptive analytics then recommends specific actions to address identified issues, optimising maintenance strategies and reducing downtime.

Tech Stack

  • Predictive and Prescriptive Maintenance Algorithms​
  • AI/GenAI​
  • Survival Analysis Techniques​
  • Anomaly Detection Algorithms​
  • Forecasting Models ​
  • Prescriptive Analytics Tools

Vehicle Health Management with AI

Aftermarket

Tata Elxsi Use Case

Use Case: Vehicle Health Management

How it Works: AI models continuously analyse data from various vehicle sensors, including engine diagnostics, performance metrics, and historical maintenance records. By identifying patterns and anomalies in this data, the models can predict potential issues such as component failures or maintenance requirements.

Based on these predictions, maintenance tasks can be scheduled proactively to address potential issues before they escalate into major problems, minimising downtime and improving overall vehicle reliability.

Tech Stack

  • Machine Learning Algorithms
  • Telematics and IoT Sensors​
  • Data Analytics Platforms

Productivity

Automation and Validation

Product Development​

Tata Elxsi Use Case

Use Case: Automotive Test Case Generation Using GenAI​

How it Works: Concepts of RAG, In-context learning, and prompt engineering is used to generate automotive test cases from system requirements. Historic requirements and corresponding test cases are given as context either using RAG or In-context learning to generate new and valid test cases.

Generated test cases shall be traceable to system requirements. The framework is also capable of generating test scripts from the test cases using the Robot framework. Thus, the GenAI framework is capable of augmenting the testing procedure, given proper context is available. This method can save time, money, and cost when compared with manual test cases, and test script generation.

Tech Stack

  • GenAI Models ​
  • RAG​
  • In-context Learning​
  • Prompt Engineering​
  • Test Case, Test Script Generation​
  • Robot Framework

Software Development Lifecycle

Product Development​

Tata Elxsi Use Case

Use Case: Requirement generation, Requirement review, Code Generation and Testing Using Gen AI for Automotive Embedded Software​

How it Works: The framework integrates Gen AI to generate requirements, review requirements, transform requirements into source code and subsequently generate test cases for comprehensive testing. By abstracting and automating the V model steps, it ensures efficient and accurate software development and validation. ​

​This is implemented for various Electonic systems in an automobile. Domain knowledge and historic artefacts are used provide proper context to the GenAI model to generate proper design and software artefacts. Inclusion of GenAI for software development lifecycle can bring in efficiency and optimisation.

Tech Stack

  • GenAI Models ​
  • V Model Software Development Process ​
  • Code Generation Tools ​
  • Test Case Generation Tools
  • Simulation and Testing Tools ​
  • Requirement Engineering