A view to build a business around SDV
Authors: Aravind D. and Prashant Gupta
Content
1????? Objective of this document?
2????? Software Businesses in Automotive ecosystem???
3????? Telematics as a Valued Service?
3.1??????? Improvement 1: Increased Personalization??
3.2??????? Improvement 2: Seamlessly integrate transactions?
4????? Vertically integrated solution for Patron Customer?
4.1??????? Building up Base Technology Stack Iteratively?
4.1.1???? Core Platform???
4.1.2???? Internal Applications?
4.1.3???? Customer Applications?
4.1.4???? Other extensions?
4.3??????? Building Additional Value-add services and extensibility?
4.4??????? Parallel Execution for Consumption and Revenue Generation??
4.5??????? Build a strong AI team???
5????? Example Opportunity:
5.1??????? Care for your Vehicle?
5.2??????? Agricultural Solution??
6????? Industry collaboration??
7????? Acknowledgement?
1 Objective of this document
This document will explore the longer-term potential and considerations for establishing a software business unit, as well as the immediate demands of internal software consumption for in-house developed vehicles. I share my perspective on a software unit's role, short and long term.?
The software need in the current automotive context has large areas of opportunities as mentioned below:
-??????? Opportunity 1: Address the needs of the vehicles built in-house. This includes building internal telematics solutions for the vehicles, improving efficiency of energy management/battery management, motor control units, interaction on the vehicle, etc.
-??????? Opportunity 2: Build new solutions to automotive ecosystem industries and adjacent markets of the vehicles built as part of “Opportunity 1”. This includes solutions for service support networks, enabling sales, building new insurance models / products, etc. This can also include solutions enabling the customers of the productivity segments who build the vehicles, farming solutions for farmers, toll management solutions for truck drivers, all regulatory assistance to an autorickshaw driver, etc.
-??????? Opportunity 3: Leveraging software expertise to generate more revenue across the ecosystem. This involves reselling/white labeling the solutions developed as part of "Opportunity 2," as well as assisting other OEMs by providing software engineering knowledge to OEMs interested in developing their own.
-??????? Opportunity 4: Scale the software team’s skills to additional business across the parent company. This comprises Industry 4.0, insurance tech, and partnership with other tech businesses in whom the parent company has invested.
We focus on Opportunity 1 and 2 in this document. As the strategy must be coherent with the parent company's overall vision and perspective, the strategy and business proposal will be "next steps" and will not be included in this document.
2 Software Businesses in Automotive ecosystem
Software skill utilization can be quantified along two dimensions.
The categorization of offering based on services, products, IP, and people is a useful way to understand what a company can provide to the market. The four segments are created by mapping the x and y axes of the chart with service, unique IP, people, and product. The fifth and sixth? segments are an ecosystem domain.
We will start with segment 4 (S4), the focus is on developing solution offerings that can be used internally and can host services to provide offerings externally. This includes creating products that can be used for value-added services, such as testing as a service, telematics services, and regulatory-driven services (like battery temperature alerting).
Segment 1 (S1) focuses on providing general SW services to the market. This segment is staffed with general skilled resources that can provide a wide range of services to the market. The services provided in this segment are not specialized, but instead, are meant to be general-purpose services that can be used in a variety of different industries.
Segment 2 (S2) is focused on creating products and services that can be sold to the external market. This segment uses the solutions developed in segment 4 to develop products based on specific specifications or white label the existing solution as is. This segment is focused on creating a variety of products that can be sold to a wide range of customers in different markets.
Segment 3 (S3) is where the company provides domain specialist services to the market. This segment is staffed with skilled resources that can provide consultancy in specific domains such as motor optimization, battery optimization, analytics, and data. This segment is also responsible for developing products that are used in the segment 4 offerings.
In addition to the four segments that represent services, products, IP, and people, there is a 5th segment & 6th segment that represents businesses in the automotive ecosystem and adjacent to the automotive segment respectively. These businesses are part of the ecosystem that the company operates in and are closely related to the company's core business.
The businesses in segment 5 (S5) include fleet management, safety and insurance, mobility and transport, aerospace, fintech and e-commerce and logistics. These businesses are unique in that they are both consumed internally and externally. This means that the company can use these services and products internally to improve its operations, and can also offer them to the external market as value-added services. By developing businesses in this segment, the company can create a more robust ecosystem that can provide a wide range of products and services to both the internal and external markets.
In segment 6 (S6) we intend to go beyond mobility and cater for the solutions that are derived out of SW services. These include the SW solutions for Aerospace, Manufacturing automation, Fintech etc.
The six segments provide a useful framework to compare and contrast how a company can create products and services that can be consumed by the internal and external markets. The focus of each segment is different, and by understanding the specific needs of each segment, a company can develop products and services that are tailored to the needs of each market.
?With the above pro/con between each segment of the software focused delivery, below is summary of comparison
To be a product-driven company, we must be in the S4 and S5 segments. The remaining sectors are focused on developing solutions for other OEMs. The primary revenue streams in the S4 includes:
1.????? Selling vehicle features: Autonomous drive, connectivity, “unlock performance mode”.
2.????? Selling subscriptions: Maps, audio/video media, trip specific packages.
3.????? Maintain resale value:? Predictive maintenance, realtime market value based on current conditions.
4.????? Data services: For automotive ecosystem industries and adjacent industries including insurance, service, component suppliers, etc.
3 Telematics as a Valued Service
Traditional Telematics has been defined as a reflection of the current and past status of the vehicle. This includes features like seeing the location of the vehicle, past trip summary, status of fuel, etc. These are “hygiene” features.? However, the customer doesn't associate too much value on reflection of their own data. There is an initial “wow” for the first time user, but a quick reduction of interest in the solutions. This has been a long lasting problem in creating enough value for the customer, to the point where they can differentiate the product using these services, and are willing to pay for these solutions.
We feel the missing ingredient here is understanding the complete usage pattern of the driver, not only when driving the vehicle, but every interaction around the vehicle. Then improve the quality and consistency of these interactions.
We will discuss 2 key points of improving the Telematics Services as part of Software Defined Vehicle.
3.1 Improvement 1: Increased Personalization
Below is an example of how the use-case of “range calculation on a vehicle” can be made more accurate to the customer, and start providing value by integrating data from the rest of the users.
?In the above range example, Level 1 is a simple formula based on battery conditions. Level 2 is personalized to the user based on their driving style. Level 3 & 4, we are adding data from the ecosystem for the benefit of the user, which can be chargeable to the user.
1.????? Level 1 is reflection of vehicle data back to the user. This kind of “reflection” use cases can’t be charged to the customer.
2.????? Level 2, we add our IP of understanding users driving style and personalizing range. We bring in some differentiation, might be chargeable, but not strong enough.
3.????? Level 3 & 4, we bring in data that the user doesn’t have access to. This level of accuracy is worth charging the user, as it will accurately allow the user to plan their day and increase their productivity.
In this process, the vehicle continues to adapt more to the user, their driving style and patterns. Making it an inherently personalized “software defined vehicle”
3.2 Improvement 2: Seamlessly integrate transactions
This is because they address any “daily usage” features. For example, instead of knowing the location of the vehicle everyday, if there is a notification of theft/unexpected activity whenever it happens, it is of better value. However, theft is not a “daily use case”. If proximity based unlocking is activated based on location (office and home), then it is a feature used daily by the user, gaining better perceived value.
When we study most features in any vehicle software, they are very focussed on allowing users to interact with the vehicle. Not much on how the features enable the daily workflow of the user. I take 2 examples in here:
1.????? Enabling parking by integrating the vehicle software with the parking infrastructure is a frequently used feature for the driver. Many of the managed parking slots in the urban areas are managed by automated parking meters. To use this, the driver has to go to a machine nearby, identify the slot vehicle is parked in, and pay. Solution: Integrate parking system with the vehicle software. By a click of a button in the vehicle, the parking reservation is done and auto extended based on the user’s needs.
2.????? Traffic fines get raised and stay in the government systems for prolonged periods of time. Many times, this is due to the driver/owner not knowing the fine was raised. The traffic fine is available in data accessible to us. Solution: integrate the Vehicle software with the traffic-fine database and notify the driver / owner everytime a traffic violation is raised by the cops.
In summary, there are 3 aspects to have users attached to a software solution
1.????? Features: Solves a key pain point to the user
2.????? Ease of use: Ability to use the features more intuitively
3.????? RAP (Reliability, Availability, Performance): Safe, Secure, Quick response.
While #2 & #3 are technical considerations, #1 needs deep insight into the problem domain from the user’s point of view. I feel this is an area that needs more insights, which are very subjective to each target segment of the users. Close coordination with the users, and fast development cycle will allow for solutions that meet the user needs.
4 Vertically integrated solution for Patron Customer
Previous section covered the shortfall of features picked by OEM providing Telematics solutions via SDV. Now we cover how we should go about building the software solution. With an assumption that there are existing solutions that are available in parts, we need to take a more pragmatic approach towards incrementally building the complete solution.
4.1 Building up Base Technology Stack Iteratively
Feature delivery will be staged to grow / evolve as the system keeps scaling up. Goal is to keep a continuous pipeline of features in the delivery pipeline, allowing us to adapt to user feedback. Below are the 4 incremental stages that can be adapted.
Side note: Before we start building, we need to understand the current baseline, which is covered in Annexure 1.
4.1.1 Core Platform
When we start building the vehicle software, we start with the Core Platform. The problems we focus on are
1.????? Energy efficiency: Power socket to battery (charging, hyper charging, …), battery to motor (discharging, thermals, …), motor to road (motor efficiency. Each of these segments of the energy lifecycle needs to have clear traceability and
2.????? Fault management: Based on various error codes, optimizations should be built to give the various user experiences.
3.????? Software updates: Ability to update software on all the vehicle components is critical for a fast turnaround on energy efficiency improvements and fault management. This helps us reduce any negative user experiences.???
4.????? Variant Management: Once we have many vehicle types using the same base software platform, we need to build the ability for software to reconfigure itself based on the type of vehicle and the experience that the customer is looking for.
5.????? Vehicle Features: All other features like cruise control, auto indicators, etc. which is envisioned.
4.1.2 Internal Applications
1.????? Manufacturability: Ability to update the software in the factory line, and keeping that efficient directly impacts the number of stations needed in the factory and hence the cost of equipment.
2.????? Supportability: Field service agents, and any other support engineers will need to have a way to make corrections / updates on the vehicle. Also the ability to diagnose / fix the problem across many vehicles becomes necessary as we scale the product.
3.????? Command center: For all of the fleet, there needs to be a way to run compliance reports, diagnosis, and solutioning from the R&D center, without needing to get physical access to the vehicle.
4.1.3 Customer Applications
1.????? Customers want to review the status of the vehicle, allowing them to monitor the health, and security of the vehicle. This includes charging status, theft/towing alerts, usage, etc.
2.????? Customers like to integrate their phone into the vehicle for calls, music, navigation and other convenience features.
4.1.4 Other extensions
With the Base Technology stack in place, the SDV has the ability to (1) interact well with the customer, (2) integrate well with the infrastructure, and (3) is ready to integrate with other software systems to exchange information to improve the services to the driver. This sets the foundation for building the next level of value added services.
4.2 Building Additional Value-add services and extensibility
This information is presented in the diagram below.
Each solution in the “Internal and External Services” needs a good reliable interface to the vehicle to make it seamless. With the foundation built in teh Platform Technology stack, it makes it predictable to integrate with the infrastructure like parking, toll management, etc. It also paves the path for these services to be extended to ICE vehicles, if the need exists.
Next step is “Eco system offering” by opening up the SDV to be a platform for other 3rd party personnel to provide services. It is enabled by providing an open API (Application Programmable Interface), SDK (Software Development Kit) and HDK (Hardware Development Kit). This will allow for other innovators to build solutions that we have not thought about, allowing us to keep in close touch with usage of the ecosystem. This will greatly benefit the user with the integrated experience for all their vehicle interactions.
So far in Section 4.1 and 4.2 we covered the aspects of what needs to be built to have an effective software defined vehicle and how to scale it to the ecosystem.
4.3 Parallel Execution for Consumption and Revenue Generation
The Development has been categorized into three phases, each with specific activities aimed at achieving the project's objectives. The activities have been further grouped into internal consumption, external revenue, automotive ecosystem industries and adjacent industry categories, depending on their relevance and focus.
The reference to S4…S5 is continued from section 2.
Phase 1 focuses on Onboarding and baselining: Objective is to understand the existing products and the roadmap for next 18 months along with its associated features. This activity will also educate about the current technologies in use and the past decisions around it. This is covered in detail in Annexure 1.? It will also include the study of the talent pool of the organization, key peoples, project mapping etc. It will also be the start of team building. The objective of this phase is to build a solid foundation for the project, set clear expectations and objectives, and create an effective team that can deliver the expected results. During this phase, the project will also conduct market research for automotive ecosystem industries and adjacent industry solution offerings. This research will help the project identify opportunities in automotive ecosystem industries and adjacent industries and design solutions that address the specific needs of these industries. All of this, while we have started building the Technology Stack needed for SDV.
Phase 2 will continue team building and begin offering general SW skilled resources to the market for training, RFQ-based delivery, and generalized solutions. The project will also create OEM spec products like white labeling of S4 products and spec-based development. The development of the platform for the products and the value-added services will begin. Additionally, strategy formulation for automotive ecosystem industries and adjacent industry solution offering will begin in this phase. By the end of Phase 2, the project should have a demo ready for one or more domains of automotive ecosystem industries and adjacent businesses, such as fleet management, insurance and policy, mobility and transportation, agriculture, and e-commerce and logistics. In this phase it’s important to execute to the strategic priorities of building S4 solutions without compromise.
Phase 3 will continue team building, platform development, and value-added services. The project will also continue offering OEM spec products and general SW skilled resources. By this time, the project should have a stable product ready in S4 and we would have upskilled a higher number of domain specialists. This makes it possible to offer domain-specific consulting, data analytics, and business transformation services. The project will create an automotive ecosystem and adjacent businesses and start generating revenue from it.
In summary, the project is designed to achieve its objectives in three phases by executing various activities grouped into internal consumption, external revenue, automotive ecosystem industries and adjacent industry categories. Each phase builds upon the progress made in the previous phase to create a solid foundation, develop value-added solutions, and generate revenue by offering services to external markets. The project also focuses on exploring opportunities in automotive ecosystem industries and adjacent industries and designing solutions that address their specific needs.
4.4 Build a strong AI team
While the overall software team is covered in Annexure 2, we need to pay attention to the 2nd wave of Software development that is changing how we develop software. AI based development is also referred to as “Software 2.0”. With AI based tools for software development, as much as 80% of the code can be auto written, allowing engineers to focus only on the simple and complex logic of the code functionality. This method of development will increase the speed and quality of development.
To enable this, we will need a strong AI team, which is not only focussed on data analysis, but internally focussed on developer efficiency and quality. AI will be the driving force throughout the vehicle value chain in the future covering design, manufacturing, transportation, sales, and servicing. Below are instances of how AI will help the complete product development lifecycle.
Design: Modern-day vehicles are engineering marvels and require a lot of sophistication. So automobile companies are quickly adopting generative designs to help them develop highly optimized vehicles. They leverage AI and ML to break down complex engineering processes and train AI-powered software to curate optimized designs based on certain parameters.
With VR, there’s no need to spend time and effort on the complicated process of building bulky physical prototypes. Virtual prototyping also simplifies research and development, speeds up the design process, and reduces the number of modifications, thus significantly cutting the costs of the entire pre-manufacturing cycle.
Software Development: AI-powered code generation tools are designed to understand the intent of the programmer and generate code that fits the requirements of the project. These tools can be trained on large codebases to recognize common patterns and identify optimal solutions. As a result, code generation tools can produce high-quality code that is well-structured and easy to maintain. They can automate repetitive coding tasks, such as boilerplate code, enabling developers to focus on more complex coding tasks. This saves developers significant amounts of time and reduces the risk of errors caused by manual coding. Additionally, code generation tools can reduce the cognitive load on developers, allowing them to focus on the overall design of the software rather than low-level implementation details. Few example tools to use
1.????? Kite: Kite is an AI-powered code completion tool that uses machine learning algorithms to predict and complete code snippets in real-time. Kite integrates with popular development environments and code editors such as VS Code, PyCharm, and Sublime Text.
2.????? TabNine: TabNine is a code completion tool that uses deep learning to predict and suggest code snippets as you type. TabNine can be used with any code editor and supports over 20 programming languages.
3.????? Deep TabNine: Deep TabNine is an advanced version of TabNine that uses a deep learning model to generate code suggestions. Deep TabNine is faster and more accurate than the original version and can be used with any code editor.
4.????? Codegen: Codegen is an open-source code generation tool that uses natural language processing to generate code from plain English descriptions. Codegen supports multiple programming languages and can generate code for a wide range of use cases.
5.????? CodeAI: CodeAI is an AI-powered code generation tool that uses machine learning to generate code snippets. CodeAI supports multiple programming languages and integrates with popular code editors such as VS Code and Atom.
Manufacturing: After Design; manufacturing process continues through the supply chain, production, and post-production. Leveraging AI in the automotive sector enables the creation of smart vehicles, as well as the use of robots and exoskeletons to make car assembly more efficient, and to improve supply chain management.
One scenario will be predicting equipment downtime. If a machine fails unexpectedly on an automotive assembly line, the costs can be astronomical. Idle employees are unable to complete their production work. AI-based algorithms can digest masses of data from vibration sensors and other sources, detect anomalies, separate errors from background noise, diagnose the problem, and predict if a breakdown is likely or imminent.
Transportation-Cockpit: AI is appearing in almost every element of car design, from driver assist technologies that reduce accident risk to monitoring systems that predict and identify maintenance needs. AR projections such as pedestrian alerts, turn-by-turn navigation, waypoint information, parking availability, charging sites, and adjacent eateries will be provided through AI-powered head-up displays (HUD). Examples of AI in the automotive industry include industrial robots constructing a vehicle and autonomous cars navigating traffic with machine learning and vision.
Autonomous Vehicle: AI is the foundation of ADAS and driverless technology, and it benefits from the development of driver assist programs, autonomous driving, driver risk assessments, and driver monitoring. Driving aid programs such as lane-change assist, parking assistance, pedestrian recognition, and so on will make driving more comfortable and enjoyable. Deep Learning, Computer Vision, Natural Language Processing and Understanding, Reinforcement Learning, Graph Neural Networks, IoTs, Cloud Computing, Edge Computing, and Causal Inference domain experts are key players in the present and future.
Edge compute: Cloud computing is utilized in cloud training and model deployment. Edge computing is employed in on-vehicle model deployment. For faster response, we can deploy non-critical and batch decision making models in the cloud. For immediate response, we can install crucial and near real-time decision making models on the edge. IoTs help to aggregate sensor, lidar, radar, and sonar data on-device and communicate it to the cloud for vehicle tracking and monitoring, as well as to improve autonomous driving.
Digital Assistant: Natural language processing and comprehension improves the vehicle-user experience. This allows the user to engage without having to touch a physical button on the car.
Sales: AI makes the purchase and selling of cars extremely simple for consumers. AI-enabled user interface systems may even recommend the best cars in the future based on the consumers' driving skills, insurance, health data, penalties received, and so on. Deep learning, machine learning, computer vision, and domain knowledge are all required. Virtual reality can be used in experience centers.
Typically, when customers start exploring the market for a new car, they need to visit a dealership. Dealers usually display a few models and often don’t even have the color customers are looking for. In such a scenario, with robust and quality VR technology, dealers are able to customize any make or model exactly the way customers are going to buy it. The purchaser will be able to walk around it, step into the driver's seat or passenger seat, and experience the vehicle.
Service: AI can be utilized for predictive and prescriptive maintenance, notifications about engine and battery performance, and insurance programs that monitor driving behavior to calculate risks and costs. Machine Learning, IoTs, domain knowledge, and causal inference are all important domain experts.
Predictive maintenance is now employed in diagnosing and correcting vehicle issues or maintenance. In the future, prescriptive maintenance will collect information from vehicles via IoT devices on a regular basis and prescribe actions/maintenance that must be performed on the vehicle.
Section 5 covers example opportunities to build as automotive ecosystem industries and adjacent segments to Vehicle sales.?
5 Example Opportunity:
5.1 Care for your Vehicle
Usually insurance programs are positioned as “care products”. In reality, they come to relevance only when there is an incident to manage. This drives the engagement with customers to be very low (once a year), and customers are focused only on cost, not value.
Alternate to that, a single window platform to buy, track, claims, assistance, service & sell via gamified & interactive customer engagement would change the mode of engagement with the customer.
Proposed Offerings:
Telematics has a full stack of data capturing options - this enables any & every use case for connected vehicles to be managed with common analytics & platform APIs.
5.2 Agricultural Solution
The use of software and digital technologies in agriculture is transforming traditional farming practices and enabling farmers to make data-driven decisions that can improve crop yields, reduce waste, and increase profits. Precision farming, farm management software, climate prediction, and robotics and automation are some of the ways in which software is revolutionizing agriculture.
India is a favorable market for agricultural tech businesses due to its large agricultural sector, increasing adoption of technology, government support, favorable climate and soil conditions, and a large population. The potential for growth and success in this sector is high, as farmers are becoming more aware of the benefits of using technology and are willing to invest in it to improve their productivity and profitability. This opens up opportunities for data-driven agriculture businesses listed below.?
6 Industry collaboration
As a new unit in the Software Defined Vehicle space, there are many ways to accelerate the journey and build uniqueness. One way is by leveraging the partner ecosystem. Partnering with established companies such as Microsoft, Intel, and Qualcomm can provide access to cutting-edge technologies, software, and hardware solutions that can be integrated into the software-defined vehicle. This could include IoT, cloud computing, artificial intelligence, and machine learning. These partnerships can provide a startup with access to expertise, resources, and networks that could be critical for their success.
Another way to accelerate the journey is by collaborating with suppliers that specialize in providing vehicle components and systems. This could include companies such as Bosch, Continental, and Magna. By collaborating with suppliers, startups can gain access to cutting-edge hardware and software solutions that can be integrated into software-defined vehicles. This could include sensors, controllers, and communication systems that are critical for the operation of a software-defined vehicle. These collaborations can help startups to build a unique product that is optimized for their target market.
Academic institutions can also be a valuable resource for startups in the software-defined vehicle space. Collaborating with institutions such as the International Institute of Information Technology (IIIT) in Hyderabad can provide access to research and development capabilities, as well as a pool of skilled professionals in the field of software-defined vehicles. Such collaborations can lead to innovations in areas such as control systems, autonomous driving, and intelligent transportation systems. By working with academic institutions, startups can stay up-to-date with the latest trends and technologies in their field and access the knowledge and expertise of academic researchers and professionals.
Engaging with bodies which define the automotive standards to influence the direction of development of products, regulations and ability to collaborate across the ecosystem. Automobile Industry Standards Committee (AISC), International Centre for Automotive Technology (ICAT), Automotive Research Association of India (ARAI), Vehicle Research and Development Establishment (VRDE),? Society of Indian Automobile Manufacturers (SIAM), Tractor Manufacturers Association (TMA)
Finally, engaging with technical communities like the TED community can be another way for startups to accelerate their journey and build uniqueness. The TED community consists of thought leaders and experts in various fields, including technology and transportation. Engaging with the TED community can provide startups with access to a global network of experts and resources that could be critical for their success. TED Talks can provide insights into emerging trends and technologies, while TED events can provide opportunities for networking and building partnerships with companies and organizations that align with the startup's vision and mission.
In conclusion, startups in the software-defined vehicle space can leverage the partner ecosystem, supplier ecosystem, and academia to accelerate their journey and build uniqueness. By partnering with established technology companies, collaborating with vehicle component suppliers, working with academic institutions, are few ways we can access critical resources, expertise, and networks that could be crucial for our success.
7 Acknowledgement
I extend my heartfelt appreciation to Saumil Kapadia and Manoj Ashokkumar for their exceptional contributions to our research paper. Their dedication and expertise significantly elevated the quality and impact of our work. Saumil's meticulous attention and Manoj's insightful analysis were instrumental. It has been a privilege to collaborate with such talented individuals, and I am grateful for their valuable input, which undoubtedly strengthened the overall success of our research endeavor.