In the world of research and development, prototypes play a huge role in testing and validation. But prototypes have their own challenges. Studies indicate that through 2022, more than 80 percent of enterprise organisations piloting AI projects will face delays in moving them to production due to lack of organisational collaboration and IT process immaturity.
By 2022, at least 50 percent of machine learning projects will not be fully deployed in production.
But what if, the role of prototypes is reduced and what if there are no prototypes at all? How about the OEMs making most of the testing even before making a prototype?You may wonder if that can be a possibility at all. This is exactly what MathWorks is trying to achieve.
The US-based software provider and developer is supporting all the four major global trends —electrification, automated driving, data analytics or predictive maintenance and functional safety standards — by providing various applications and solutions. With the evolving EV market in the Indian market, the software provider also has products for electric vehicles and its other components.
Pointing out an example, Vijayalayan R, Manager, Auto Industry and Control Design Verticals Field Application Engineering, MathWorks, mentioned that engineers at Ather Energy designed and optimised the scooter, its control software and charging stations using Model-Based Design with MATLAB and Simulink. The Ather team wanted to make it easy for owners of petrol-powered scooters to switch to the 450. The engineering team faced numerous unknowns, including rider usage patterns and they needed to run simulations to evaluate design concepts for a variety of riding and usage scenarios and make informed trade-off decisions as they optimised design parameters. In addition to designing the scooter itself, they needed to develop and implement embedded control algorithms for battery charging, temperature management, and other key functions.
The Ather team began by building a plant model of the scooter and its main components. Vehicle dynamics and mechanical components were modelled directly in Simulink using first principles. The power converter and electrical circuitry were modelled using Simscape and Simscape Electrical. In the absence of detailed component data, the team took an empirical approach to modelling the battery cells. They tested the battery at various temperatures and state-of-charge levels and used the measured input-output data with System Identification Toolbox to create a black-box model of the cell’s electrical and thermal characteristics. They developed algorithms for battery charging, power control, and temperature control in Simulink. They modelled the control logic in Stateflow and used the Control System Toolbox to tune controller gains.
“We interact with various customers and continuously enhance existing products. If required, we also come up with new products. This happens on a constant basis and with every new release, we try to add new capabilities. In the latest release of 2020b and the release before, we have added more reference examples for EVs,” he said.
Vijayalayan explained, “From the electrification perspective, OEMs and Tier 1s have their own ideologies and focus on various needs. They might either develop an entire system or a separate subsystem. From our side, we are providing the platform for them to develop an EV or the key components of the EV. Our offering will be around the system simulation to enable the engineering in EVs. This is mainly to ensure the design works even before the physical prototype and evaluate a variety of configuration and come to a conclusion.”
Vijayan highlighted that these product designs can also be reused, “The other new product for EVs is the Motor Control Blockset for the electric motors. The blockset provides a reference application for the e-motors along with algorithms and embedded targets. Then, Simulink is used in batteries, BMS and even to understand the model and behaviour of the battery. Also, using Simscape you can develop control systems and test system-level performance. Engineers can also reuse the products designed, which saves time.”
The software company releases its updated software and products every six months. It constantly keeps adding capabilities to the existing software giving more value to its customers. Major products include MATLAB and Simulink, which support data analysis and simulation. During the first launch of 2020, MathWorks introduced the Motor Control Blockset, which is an add-on product for Simulink for designing and implementing motor control algorithms. With these, engineers can use reference examples and Simulink blocks for developing field-oriented control algorithms to spin brushless motors.
Also, the Powertrain Blockset provides fully assembled reference application models of automotive powertrains, including gasoline, diesel, hybrid, and electric systems. Powertrain Blockset provides a standard model architecture that can be reused throughout the development process. Engineers use it for design trade-off analysis and component sizing, control parameter optimisation, and hardware-in-the-loop testing.
A step into autonomous
Mathworks also has products in the autonomous system space. According to the company, the term autonomous can cover multiple industries from robots to automated driving. The new release in 2020 has a UAV toolbox, which helps in drone and UAV applications. They have also strengthened their automated driving space with a new Lidar toolbox.
“Camera, radar and Lidar play a major role in an automated driving system and products will help in making these by providing algorithms. We try to provide solutions where some tasks are difficult to recreate and dangerous to make in the real world. Nowadays, ADAS is gaining more prominence and we help by creating a virtual ground for developing and testing. We are continuously investing and developing a platform for it. In India, the question is what level of autonomous are the OEMs working towards. It is about the safety of the driver and the pedestrian. We are witnessing an increase in penetration of autonomous vehicles in the automobile field and we have products like Vehicle Dynamic Blockset and Road Runner, which automates the creation of road networks from HD Maps to support it,” said Vijayalayan.
The move towards digitalisation
The recent global outbreak of the coronavirus has pushed for many new needs. The software maker also thinks the pandemic will give digitisation a major push.
“In the recently held MATLAB Expo, the theme was around the four megatrends. But, there were keynotes on pragmatic digital transformation too. Usually, we focus on artificial intelligence and its evolution. But this year, we wanted customers to make use of the digital transformation. Though the move towards digitalisation was there for a long time, engineers now look for smart methods. And, with the new megatrends and to develop that, we see more usage now. To collect data from various sources, analysing it and getting insights from it is a big challenge. And to develop an algorithm according to the challenge is even more difficult. Also, in predictive maintenance, we need to understand even before something is going to fail. The industry needs to move accordingly to support these,” he concluded.
This feature was first published in Autocar Professional's November 1, 2020 issue.