Autonomous Cars – based on end users’ needs.

From automation to fully autonomous cars. With its extensive knowledge of the automotive industry, Semcon has the resources and expertise to develop effective autonomous solutions, always focusing on the most important aspect – end users.

Autonomous solutions are nothing new. They are based on automation developed over more than a century of industrialisation, up to the robotics solutions of today. Autonomous vehicles are not new, either – they are based on Automotive Active Safety and Advanced Driver Assistance Systems (ADAS), which have been in development since the 1990s. But what are new are fully autonomous vehicles involving what is known as High Autonomous Driving (HAD), correlating the SAE Autonomous Driving Scale – levels 4 and 5.

People before technology

Semcon has been working with state-of-the-art development of Active Safety and ADAS for more than a decade alongside our leading customers, which include Volvo Cars, BMW, Volvo Trucks and Autoliv/Veoneer. We have learned many key lessons over this decade and built the foundation for HAD.

Making autonomous cars user-friendly

In the field of automotive development, another key lesson learned is that it is necessary to develop systems that help users, not worry or disrupt them. Real world investigations show that more than 50% of drivers actively disable key ADAS functions, life-saving systems such as Forward Collision Warning and Lane Keeping Assist.

Such situations could be avoided by developing systems with end users in mind. The fact that drivers disable these safety systems clearly demonstrates that these systems are not user-friendly. There is a lot of work to be done in respect of UX as regards autonomous solutions. We have a specialist UX department at Semcon in Göteborg that works with these kinds of issues.

Technical challenges within autonomous

Besides working on the UX perspective, we also face a number of key technology challenges that still need to be addressed so as to implement fully autonomous vehicles successfully. In the main, these are:

• Environmental perception: the vehicle must be able to understand its entire environment at all times. Sensor Fusion – handling information from multiple sources such as camera, radar and lidar sensors, geodata and V2X data is key. These input sources, as well as data handling and interpretation, are still being developed.
• Decision-making: a fully autonomous vehicle needs to be able to handle an infinite number of scenarios in the real world. Hence a classic code programming approach is impossible (it is not possible to hard code actions to cope with an infinite number of cases). This is why smart learning algorithms are needed.

Semcon specialises in machine learning and deep learning technologies, working with handling environmental perception and decision-making challenges. Semcon offers its customers end-to-end AI/machine learning solutions; from defining use cases with customers, to machine learning/deep learning development and finally product integration, verification and validation.

Find out more about Semcon’s work in the field of AI.

Talking Autonomous

Talking Autonomous is a series of short talks where our experts talk about different topics with regards to the future for autonomous cars. Available episodes below:

• Keeping humans in mind when developing autonomous systems
• Cognitive load, hot potatoes and hand over time
• Adapting technology to humans and not the other way around
• The challenge of validating autonomous vehicles
• Autonomous decision making
• Backseat drivers and annoying autonomous systems
• The hard work of validating autonomy