Every decision made in automotive HMI design has a consequence that no other product category shares. A poorly designed app frustrates. A poorly designed HMI endangers.

That is not a dramatic statement. It is the reality that every automotive and mobility product leader is responsible for the moment a vehicle leaves the production line. The interface between the driver and the machine is not a feature. It is the primary relationship through which every journey is experienced, every decision is made, and every trust is built or broken. 

The vehicles that drivers genuinely trust are not the ones with the largest screens or the most connected features. They are the ones where the interface feels so considered, so natural, and so invisible that the driver never has to think about it. Their attention stays where it belongs, on the road.

That standard is harder to achieve than it sounds. And the automotive HMI design principles that produce it are worth understanding before the first design decision is made.

Safety hierarchy: the foundation of automotive HMI design 

The most critical information is always the most accessible. Everything else is secondary. 

This sounds self-evident. In practice it is one of the most frequently violated principles in vehicle HMI design. 

The pressure to surface more features, more connectivity, and more content in the driver’s field of view is commercial and real. Buyers respond to specification sheets. Screens sell cars in showrooms. But the driver traveling at speed on a motorway has a hierarchy of needs entirely different from the buyer standing in a dealership. 

Speed, navigation, fuel or charge state, and hazard warnings. These are the four categories of information a driver needs to access without cognitive effort at any moment. Everything else, media controls, connectivity status, and comfort settings, should require a deliberate decision to access rather than competing for attention passively. 

When Robosoft designed the HMI for the Ford GT, one of the most performance-focused supercars ever built, the safety hierarchy principle was not a constraint. It was the brief. Every element of the interface was evaluated against one question: does a driver at speed need this, or do they want it? The distinction between need and want determined every placement decision. 

The result was an interface that Le Mans race drivers could trust at speeds where a moment of distracted attention is not recoverable. That is the standard safety hierarchy is designed to meet. 

Information density: an HMI design best practice often overlooked 

More information is not better information. The right information at the right moment is. 

The instinct in complex vehicle systems is to surface everything the system knows. The vehicle has sensor data, connectivity data, navigation data, performance data, and comfort data all available simultaneously. The temptation is to make all of it visible. 

The result, when that temptation is not resisted, is an interface that creates cognitive load rather than reducing it. A driver who must scan a dense display to find the information they need at a specific moment is a driver whose attention has been diverted from the road to the screen. 

Great automotive HMI design is as much an act of subtraction as addition. Every element that appears on a display should have earned its place by being the most useful thing that could appear at that moment in that context. Context is the key word. The information a driver needs at the start of a journey is different from what they need at motorway speed, which is different again from what they need in heavy urban traffic. 

When Robosoft designed the HMI for the Triumph motorcycle, a full colour circular display featuring two unique modes and multimodal Information on Demand control, the information density principle shaped every mode decision. The right information surfaces at the right moment. The rest waits until it is needed. The result earned the HMI Europe Award and put over 150,000 units on the road with an interface that riders trust at speed. 

Mode awareness in automotive cockpit design

The driver always knows what state the vehicle is in. No surprises. No ambiguity. 

Modern vehicles are complex systems with multiple operating modes, drive modes, regenerative braking settings, autonomous features, connectivity states, and charging configurations. The number of states a vehicle can be in has grown significantly as software becomes the defining layer of the modern vehicle. 

The HMI challenge that complexity creates is significant. A driver who is uncertain about what mode their vehicle is in, or who does not know that a mode has changed, is a driver who cannot fully trust their vehicle. That uncertainty, even when it never produces an incident, erodes the relationship between driver and machine over time. 

Mode awareness is the principle that ensures the driver always has a clear, immediate, unambiguous understanding of what state the vehicle is in. Not buried in a menu. Not requiring interpretation. Visible, clear, and immediately understandable. 

When Robosoft designed the automotive cockpit design for the Volta Zero electric truck, mode awareness was a primary design challenge. The Volta Zero features a triple screen cockpit built for a central driving position, an unconventional configuration that placed the driver at the centre of a complex information environment. In an urban delivery vehicle operating across multiple modes throughout a working day, the driver needs to know the vehicle’s state at any moment without having to look for it. The interface was designed to make that knowledge effortless. 

Emotional response: where HMI design best practices meet human experience 

HMI is not just functional. It is the primary sensory relationship between a driver and their vehicle. It must feel right, not just work right. 

This is the principle that separates good HMI from great HMI. Every other principle on this list is a functional standard. This one is a human standard. 

Drivers do not describe their relationship with a vehicle in functional terms. They describe how it makes them feel. Confident. Connected. Exhilarated. Safe. In control. These are emotional responses, and the HMI is one of the primary inputs that shapes them. 

The emotional response principle asks a different set of questions from the others. Not only does this information appear in the right place, but does this interaction feel satisfying? Not is this mode clearly communicated but does changing modes feel purposeful and considered? Not does this system boot quickly, but does the startup sequence feel like the vehicle is coming alive? 

These are questions about craft. And they require the same rigour in their answers as any functional requirement. 

The Triumph motorcycle HMI was designed with this principle at its centre. A motorcycle rider’s relationship with their machine is one of the most emotionally significant relationships in personal transport. The HMI had to honour that relationship, not just functionally, but in every interaction, every animation, every moment of feedback between rider and machine. That emotional consideration is part of what put 150,000 of them on the road. 

What these automotive HMI design principles mean for mobility leaders today

The electrification and connectivity of vehicles is creating more HMI design decisions, made more rapidly, than at any previous point in the industry’s history. Every new platform, every new vehicle type, every new feature set requires HMI decisions that will shape the driver’s experience for the lifetime of that vehicle. 

The principles on this list are not a checklist. They are a way of thinking about the relationship between a driver and their vehicle’s interface that produces decisions that hold up across every vehicle type, every operating condition, and every driver profile. 

The Ford GT proved them at Le Mans. The Triumph proved them at 150,000 units on the road. The Volta Zero proved them in the specific demands of urban commercial delivery. Three completely different vehicles, three completely different briefs, one consistent standard. 

That consistency is not accidental. It is what principled automotive HMI design produces. 

Robosoft Technologies designs and engineers HMI systems for some of the world’s most demanding automotive and mobility clients, from award-winning motorcycle interfaces to Le Mans-winning supercar cockpits to next-generation electric vehicle platforms. If you are thinking seriously about HMI design for your next vehicle program, we would be glad to have that conversation.

Self-driving cars rely on machine learning (ML) algorithms that require large volumes of accurately labeled data to function safely and reliably. Most ML models used in the automotive industry today are supervised, which means they must be trained on annotated datasets that define and categorize real-world driving scenarios. 

Data annotation enables autonomous vehicles to understand their surroundings, identify lanes, detect objects, and make split-second decisions. In the sections ahead, explore data annotation for autonomous vehicles and how labeled data supports the training of reliable AI models.  

Just like a child who must first grasp basic concepts like circles, squares, and triangles to recognize shapes before diving into complex patterns, machines follow a similar learning curve. Machines require a way to decode and understand the data they receive. This is where data annotation comes in. 

What is data annotation? 

Data annotation is the practice of adding descriptive labels, tags, or classes to raw data so that a machine-learning model can understand what it is seeing, hearing, or reading.  

In the context of autonomous driving, consider an image-recognition scenario where a folder of unlabeled street photos is nearly useless to an algorithm. Once each image is tagged with “pedestrian,” “traffic light,” or “truck,” the dataset gains meaning. The model can now associate patterns with real-world objects and predict them in new scenes.  

What does data annotator do? 

A data annotator plays a key role in preparing datasets for machine learning by labeling elements within raw data. This involves drawing bounding boxes around objects, highlighting regions of interest, tagging attributes, or classifying entities across various data types, such as images or LiDAR scans. 

For example, annotators manually review thousands of frames from dashcams or street cameras to train a model to detect traffic lights, labeling each instance of a traffic light with precision. 

Key object categories annotated for self-driving systems include: 

• Vehicles (cars, trucks, motorcycles) 

• Cyclists 

• Pedestrians 

• Traffic signs and signals 

• Lane markings 

• Road hazards like potholes, barriers, or construction zones 

Types of annotations for autonomous vehicles 

Accurate data annotation for autonomous vehicles depends on two primary sensor streams, camera images and LiDAR point clouds. Each stream demands specialized techniques to extract the detail an autonomous driving stack needs. 

Image Annotation

Image annotation is the most common form of data annotation used in computer vision. It involves labeling objects or areas of interest within 2D images captured by vehicle-mounted cameras. Key image annotation methods include: 

• Bounding Boxes: a widely used annotation method, bounding boxes are rectangular frames drawn around objects to define their position and dimensions clearly. This technique is extensively applied in object detection, training models to recognize pedestrians, vehicles, traffic signs, and other critical objects.

• Semantic Segmentation: it provides pixel-level annotation by assigning each pixel in an image a specific class label, such as road, sky, vehicle, or pedestrian. This detailed annotation enables autonomous systems to understand scene composition and make nuanced decisions, especially in high-density traffic or complex urban environments. 

• Instance Segmentation: unlike semantic segmentation, which classifies object types, instance segmentation identifies and outlines individual objects, even when multiple items belong to the same class. For example, it distinguishes between two cars or pedestrians walking side-by-side. This is crucial for systems that can track and react to separate moving objects independently.

• Polygon Annotation: it allows annotators to define object boundaries using flexible, multi-point shapes—ideal for irregularly shaped or overlapping objects like trees, traffic islands, or curving lane markings. It delivers greater accuracy than bounding boxes in scenarios where object contours are critical. 

LiDAR Annotation 

LiDAR (Light Detection and Ranging) technology emits laser pulses to measure the distance and shape of objects, generating detailed 3D representations known as point clouds. Annotating LiDAR data enables self-driving cars to accurately perceive depth, location, and shape in three-dimensional space. 

Key characteristics of LiDAR annotation include: 

• LiDAR Point Clouds: LiDAR scanners collect vast datasets containing billions of individual data points, each with precise coordinates and reflective attributes. Annotators label these point clouds to distinguish vehicles, pedestrians, cyclists, road signs, and other critical features, facilitating robust spatial understanding and precise navigation. 
• 3D Bounding Boxes (Cuboids): Similar to their 2D counterparts, 3D bounding boxes encapsulate objects within a three-dimensional space, providing essential information about their size, orientation, and location. These annotations are crucial for reliable object detection and accurate real-time obstacle avoidance. 

Best data labeling tools 

Selecting the right annotation platform accelerates your data annotation for autonomous vehicles workflow and ensures consistency across large teams, whether in-house or through data annotation outsourcing. 

• CVAT: an open-source tool by Intel, offering bounding box, polygon, and semantic segmentation support for images and videos. 

• Labelbox: a cloud-based platform with collaborative features, customizable workflows, and built-in QA modules for scalable annotation. 

• VoTT (Visual Object Tagging Tool): a free, Microsoft-developed desktop app optimized for bounding box and rectangle labeling, with Azure integration. 

• Xtream1: a comprehensive, enterprise-grade solution supporting multi-sensor (image, LiDAR) annotation and automated quality checks.
• CloudCompare: an open-source tool specialized for 3D point-cloud annotation, ideal for precise LiDAR labeling tasks in autonomous driving. 

Your annotation team should also be proficient with any client-specific tools or proprietary platforms to maintain alignment with existing workflows and data formats. 

Data labelling best practices 

• Establish detailed annotation guidelines with standardized label definitions and update them regularly to cover new driving scenarios. 

• Define a clear workflow → annotation, quality review, rework, final review, and delivery – with assigned roles and deadlines. 

• Maintain a continuous feedback loop with clients to align on labeling expectations and minimize rework. 

• Provide ongoing, project-specific training for annotators to ensure consistent understanding of tasks and tools. 

• Perform regular manual QA checks using predefined checklists to identify and rectify annotation errors early in the process. 

• Implement version control for annotated datasets to track changes, support audits, and simplify error tracing. 

• Start with a pilot annotation phase on a representative dataset to refine guidelines and workflows before scaling up. 

Our proven end-to-end data annotation workflow 

Our data annotation services enable automotive clients to build high-quality datasets by following a well-defined, end-to-end workflow. 

1. Tool selection: we begin by choosing the optimal annotation platform for the project’s needs. The team evaluates whether to use industry-leading tools like Labelbox or CVAT, or a client’s in-house toolset. By aligning the tool choice with project requirements (e.g. support for 3D point clouds or specific export formats), we ensure smooth integration into the client’s development pipeline.
2. Dedicated data annotation team: then we choose a dedicated team of experienced annotators and project leads who understand the nuances of automotive data. The team studies the client’s labeling guidelines in detail. Robosoft also provides thorough training sessions and even conducts a pilot project on a small data subset to confirm that guidelines are correctly interpreted.
3. Semi-automated acceleration: where appropriate, pre-trained models generate initial labels for rapid review, and human annotators then correct and enhance these predictions to meet accuracy targets.
4. Efficient annotation execution: throughout the annotation execution, we emphasize accuracy and consistency. Annotators follow the agreed-upon guidelines strictly, and complex edge cases are flagged for discussion so that the entire team handles them uniformly.
5. Multi-level Quality Assurance: the process includes multi-level reviews of the annotated data. Senior annotators perform second-pass reviews on samples to catch any errors or inconsistencies. Any discrepancies are corrected, and feedback is looped back to the annotation team to prevent repeat issues.
6. Delivery and ongoing support: final datasets are exported in the client’s preferred formats (COCO JSON, XML, PNG masks, etc.) and delivered in iterative batches to match development sprints. Post-delivery support includes update patches, re-annotation for new scenarios, and performance reviews. 

Fueling the future of SDVs with data annotation services 

The move to fully autonomous vehicles starts with understanding what is data annotation and ensuring every object, lane, and signal is labeled correctly. Without this foundation, AI models cannot achieve the precision required on today’s roads. 

At Robosoft Technologies, we’ve partnered with leading OEMs to accelerate their transition to the Software Defined Vehicle (SDV) market. Our expert teams deliver accurate and secure data annotation services, leveraging advanced labeling techniques, best-in-class tools, and thorough quality checks. This ensures your autonomous driving projects receive the high-quality annotated data necessary to build reliable AI models and drive future innovation.

The automotive industry is pivoting to a transformative phase resulting from a shift to electrification, tighter emission regulations, and the growing complexity of vehicle software. By 2030, the share of electric cars is set to surpass 40% in overall car sales. So, automotive OEMs and Tier 1 suppliers experience increased pressure to deliver optimized, robust EV (electric vehicle) powertrain systems faster and at less cost. Consequently, optimizing automotive powertrain systems has become a critical process that organizations must keep in check to enhance vehicle performance and ensure compliance with stringent environmental regulations.

Model-Based Design (MBD) is an engineering methodology in which mathematical models are used to design, simulate, verify, and validate the behaviour of powertrain systems before hardware implementation. It enables the use of digital twins on top of integrated simulations to facilitate iterative design and testing in a virtual environment.

This article explores how Model-Based Design changes automotive powertrain development, delivering measurable ROI to engineering organizations while supporting long-term digital transformation initiatives like Model-Based Systems Engineering (MBSE) and digital twins.

The strategic imperative of Model-Based Design for automotive powertrain systems development 

The powertrain—comprising the internal combustion engines (ICE), transmission, and drivetrain—is the heart of any vehicle. Optimizing its performance is a balancing act involving emissions compliance, fuel efficiency, cost control, and consumer satisfaction. Traditional development methods rely heavily on physical prototypes and sequential testing and are increasingly inadequate. They are slow, expensive, and prone to late-stage discovery of faults.

Model-Based Design is an engineering methodology in which mathematical models are used to design, simulate, verify, and validate the behaviour of powertrain systems before hardware implementation. It enables the use of digital twins and integrated simulations and facilitates iterative design and testing in a virtual environment.

At the core, it involves: 

• System modelling and architecture definition: Simulating various interactions of whole powertrain components.

• Plant modelling: Engine thermodynamics, motor behaviour, battery systems.

• Control algorithm design: Torque control, gear shift logic, energy management.

• Simulation & testing: Test and refine models by employing Model-in-the-Loop (MIL), Hardware-in-the-Loop (HIL), and Software-in-the-Loop (SIL) frameworks.

• Verification and validation: System behaviour validation of powertrain design under various conditions for safety compliance like ISO 26262.

• Automatic code generation: Automatic generation of production-ready code from system models for Electronic Control Units (ECUs) production using tools like MATLAB/Simulink and dSPACE.

Components of Model-Based Design for automotive powertrain systems

• System Modelling
  Model-based workflows start with system modelling at the level of system dynamics of the whole powertrain. These models are used in shared environments like MATLAB/Simulink, allowing engineers to simulate the interactions of various components like the engine, transmission, and hybrid systems with one another. This system-based approach will help identify bottlenecks and optimize performance under different operating conditions.
• Control Algorithm Development
  Robust control systems are the key to achieving high performance in the powertrains. MBD allows teams to design, test, and refine the control algorithms in simulation. For example, calibrating the engine control units for enriching combustion efficiency or the battery management system for electric vehicles could be carried out without physical prototypes.
• Automatic Code Generation
  The MBD process generates production-quality code to translate design models using tools like Embedded Coder, which helps automate the development process. A significant advantage of this is supporting AUTOSAR-compliant architectures for ECU integration, minimizing errors, and accelerating functional deployment to ECUs.
• Simulation & Validation
  Powertrains operate on different physical phenomena, which extend to thermal, mechanical, and electrical dynamics. MBD platforms integrate these domains to assess trade-offs and ensure that designs meet efficiency standards, durability, and regulations. The model behaviour is tested in several frameworks, such as MIL, SIL (software-in-the-loop), and HIL, to achieve model validation through the stages of development.

Case in point: Realizing efficiency in hybrid powertrains

Consider a manufacturer that designs and builds a hybrid electric powertrain. A traditional method would likely involve constructing several prototypes to evaluate different power distribution strategies, thermal management, and emissions compliance.

With MBD:

• Engineers can model the whole hybrid system virtually and simulate different driving conditions.

• They can digitally design and validate control strategies for energy management.

• Optimizing thermal performance with multi-physics simulations guarantees that the battery will last longer and consume less energy.

• HIL testing helps them validate the integration of highly integrated software and hardware.

This approach reduces the need for physical testing and enables the rapid exploration of “what-if” scenarios, such as varying battery sizes or electric motor configurations.

Implementation challenges and strategic considerations

The advantages of MBD are apparent, but its implementation involves overcoming both organizational and technical hurdles.

• Skill Gaps: Effective adoption of MBD heralds the upskilling of team members in workflow and simulation tools.

• Toolchain Integration: Organizations must ensure seamless integration between MBD platforms and existing tools.

• Cultural shifts: Transforming from prototype-driven to simulation-based development demands cultural buy-in from all stakeholders.

Top executives play a significant role in overcoming this barrier. Investment in training, provision of cross-disciplinary collaboration, utilization of offshore engineering, and flexible, scalable MBD tools will ensure a smooth transition.

A competitive imperative

Model-Based Design stands out as an engineering methodology and a competitive differentiator in the ever-increasing competition for intelligent, green, and software-oriented vehicles. It allows automotive companies to bridge the gap between concept and production to innovate powertrain systems with high safety or compliance.

Simulation and automation embedded at the centre of powertrain development offer transformational value to engineering leaders through reduced costs, faster delivery, and futureproofing of their mobility systems for the next decade. As a trusted partner in Embedded Engineering solutions, we enable Automotive OEMs, innovators, and Tier 1 suppliers to harness the power of simulation for high-performance powertrain design innovation.

Cars have long conveyed basic information to drivers, like a warning light for low fuel or a door-ajar alert. But Vehicle-to-Everything (V2X) communication takes this a step further by enabling real-time “conversations” among cars, pedestrians, bicycles, road signs, and traffic signals.

The result?

• More energy-efficient driving
• Enhanced safety
• Better user experiences

However, building these complex systems demands an innovative approach, one that Model-Based Design (MBD) with MATLAB and Simulink delivers exceptionally well.

Vehicle-to-Everything (V2X) communication

V2X communication is critical for achieving:

• Safety: vehicles can instantly share alerts on road hazards, emergency braking, or adverse weather conditions, thereby reducing collisions.
• Traffic efficiency: integrating vehicles with infrastructure (V2I) helps manage traffic signals, easing congestion in busy urban areas.
• Environmental benefits: smarter route planning and efficient driving patterns cut fuel consumption and emissions.
• User experience: through Vehicle-to-Network (V2N) integration, drivers benefit from advanced infotainment, navigation, and cloud-based services.

To ensure these benefits, V2X systems must reliably manage data exchange in shifting road environments, adhere to stringent industry standards, and integrate seamlessly with the vehicle’s Electronic Control Units (ECUs).

Why Model-Based Design for V2X?

Model-Based Design employs models as the central artifacts throughout the development cycle. From concept and simulation to deployment and verification.

MATLAB and Simulink offer a robust toolkit for V2X development because they enable:

• Rapid prototyping: engineers can quickly design V2X algorithms and test them in virtual environments.
• System-level simulation: simulink helps integrate vehicle dynamics with V2X models, facilitating holistic evaluations of communication and control strategies.
• Code generation: automated production of C/C++ code ensures that validated models can be seamlessly deployed to embedded hardware.
• Standards compliance: built-in support for IEEE 802.11p (DSRC) and cellular V2X (C-V2X) helps meet regulatory and interoperability requirements.

Development workflow for V2X using MBD

Requirements analysis

• Define system requirements based on V2X standards and real-world use cases (e.g., collision avoidance, traffic light prioritization).
• Track and manage requirements using tools like MATLAB Requirements Toolbox.

System modeling

• Create high-level models of V2X communication protocols, such as DSRC or C-V2X, in Simulink.
• Model application layers (e.g., cooperative awareness messages) and physical layers (e.g., radio wave propagation).

Simulation and testing

• Validate system behavior under realistic conditions (e.g., urban intersections, highway platooning).
• Employ MATLAB’s communication toolboxes to replicate wireless channel impairments, ensuring robust performance in variable environments.

Code generation and integration

• Automatically generate optimized C/C++ code using Simulink Coder or Embedded Coder.
• Integrate the generated code with ECUs for thorough in-vehicle testing.

Hardware-in-the-Loop (HIL) testing

• Evaluate the system on actual hardware in real-time, verifying performance across various operating conditions.

Deployment and validation

• Deploy the final V2X system onto the vehicle’s hardware platform and conduct on-road testing to confirm reliability and compliance.

Realizing the benefits

By embracing MBD with MATLAB and Simulink, automotive engineers can:

• Reduce development time: early detection of design flaws through simulation significantly cuts down on late-stage rework.
• Enhance reliability: rigorous testing at each stage fortifies overall system resilience.
• Streamline collaboration: unified models create a shared, accessible framework for diverse teams spanning software, hardware, and systems engineering.
• Facilitate scalability: modular structures make it easier to adapt V2X solutions for different vehicle classes and communication technologies.

Conclusion

As connected vehicles continue to redefine modern transportation, Vehicle-to-Everything (V2X) communication stands at the forefront of safer, more efficient mobility. Leveraging Model-Based Design with MATLAB and Simulink enables engineers to navigate the inherent complexities of V2X development with greater precision, ensuring these advanced systems meet the evolving demands of tomorrow’s automotive landscape.

Developing effective V2X solutions for connected vehicles requires not only a solid command of Model-Based Design but also in-depth expertise in automotive engineering. At Robosoft, we combine our extensive automotive experience with Model-Based Design (MBD) capabilities to streamline the entire V2X development lifecycle. Explore our Core Engineering solutions to bring your connected mobility vision to life.