Drawing on decades of technology expertise from across the Volvo Group, the newly formed Volvo Autonomous Solutions (VAS) has been established to accelerate the development, commercialization and sales of autonomous transport solutions. This presentation explains how the combined force of the group’s dedicated vehicle technology is driving customer solutions on and off road. From the group’s first self-driving mining truck and 5G-operated remote-controlled wheel loader through to the world’s first emission-free quarry and now to the commercial pilot of an autonomous electric hauler, VAS is leading the industry with a total autonomous transport package.

This review will present recent advances in the field of off-highway autonomous driving, including the technologies our team is developing to ensure smooth and comfortable off-highway autonomous driving, while ensuring that these technologies are robust in all weather conditions. The speaker will also present some of the technologies the team has been working on for advanced analytics using AI.

Danfoss shows that it is an innovative company in autonomous machines by developing solutions for off-highway applications. The goal is to provide machine platforms with advanced capability using Danfoss products that empower customers to build machines with autonomous capabilities using sensors, operating in unstructured cluttered environments, and collaborating with human operators. The Danfoss autonomous skid steer loader has lidar scanners, IMU and a powerful onboard computer built by Danfoss Power Solutions. The software solutions in Danfoss’s PLUS+1 autonomous control library empower OEMs and system integrators to create modular systems that can scale to multiple vehicle applications.

Driven by the need to increase productivity while improving operator safety, the development of autonomous operations and advanced control systems is spreading quickly within all industry market areas (agriculture, construction, material handling and mining). Indeed, automating part or full operation is a solution that most of our customers are looking at or are already investigating. During this presentation, Siemens will present its vision on the four key pillars leading to a strong simulation framework dedicated to machine autonomous operations virtual development: environment, physical sensors, controls and multi-disciplinary digital twin. The presentation will also show a virtual demonstrator for agricultural application, used to accelerate ADAS design and control validation.

Yeti Snow Technology and Øveraasen have won the world’s first commercial contract for delivery of autonomous snow removal equipment at airports. Over the next few years, the Swedish airport operator Swedavia will start using unmanned vehicles for winter maintenance of runways and taxiways at its airports. The presentation will cover the journey from the first vision of autonomous snow removal at airports to implementation at a major international airport in Scandinavia.

ASI has been developing autonomous industrial vehicles for 20 years in markets including agriculture, mining and construction. The presentation will share the painfully learned lessons on that journey and case studies of projects in which they were learned. These lessons will touch on topics including hardware, software, system integration, testing, validation, change management and commissioning.

Autonomous machines in agricultural applications are rapidly gaining momentum as early systems are launched and OEMs are developing systems for many different applications. This is driven by the need to increase food production for a growing population, but with a shortage of skilled labor, increased demand for sustainable farming methods, and limits reached for machine size. These autonomous machines require technologies from robotics industries to be integrated with rugged machine control system technology suitable for agricultural mobile machines. This talk will describe some of the common areas that are the biggest challenges in this integration and different approaches that have been taken to overcome these, drawing from examples of real-world autonomous machines to illustrate.

The session discusses three autonomous solutions in the farm, mining and construction equipment industry.The requirement for an autonomous solution package as an add-on for driverless endurance testing of a tractor in muddy terrain arose from the pandemic situation and thoughts on automating endurance testing and remote monitoring. The tractor has to move repeatedly in specified patterns for days. The solution developed as an add-on kit would help the OEMs in driverless-testing vehicles in a managed environment.The autonomous sand loader is thought of in view of the health of operators. Aligning the loader vehicle to the receiver truck, distributing material in the truck’s dump area, and path planning in the unstructured environment are the key challenges.Autonomous tractors for farming applications is another requirement discussed. Guidance and localization in the unstructured environment and precision control are significant challenges in this sector. Various localization approaches are discussed in this session.

A fundamental challenge in the development of autonomous machines is how to generate quality training and test data more quickly and cost-efficiently. rFpro brings the real world to simulation, where our highly accurate digital models provide a realistic environment for training, testing and validating autonomous vehicles and ADAS systems. We also enable developers to expand their data production and deliver synchronous simulation across multiple vehicle sensors through an approach called data farming. This method can successfully support manufacturers of industrial vehicles to improve their autonomous products.

ADAS and AD applications are as important in off-road as they are in on-road – perhaps even more so. Although given less press attention, off-road applications benefit from confined areas to make fast progress in autonomous applications. The requirements for the sensor suite for off-road match most of the on-road requirements, with more importance placed on shock and contamination resistance. In some cases, such as mid-summer harvesting in Arizona, the functioning of the equipment depends on it, with or without autonomous driving. The paper describes the specific off-road requirements for lidar and discusses use cases and tests to validate the performance.

Mobile machines are increasingly moving toward automated and autonomous driving. The requirements of developers in test and validation for AD functions are on the rise. The presentation covers the main technical obstacles in dealing with multi-sensorics – focused on camera, radar and lidar – including systems and tool requirements, and gives an insight into best practices to overcome the challenges. Topics range from sensor data extraction via the recording of multiple sensor data streams, to the ingest into a data center or cloud infrastructure for further analysis, function development and AI training.

Thanks to the emergence of the self-driving car, lidar hardware technology has experienced tremendous progress. However, for this technology to reach industrial applications beyond automotive, it needs to be able to deliver full situation awareness in complex off-road situations. In this presentation we will discuss the role that specific real-time lidar processing software will play in this transition.

In order to trust an industrial vehicle to drive autonomously in a safe way, there is a need to train the software perception stack and then to validate its performance versus ground truth. This presentation will demonstrate how machine-assisted labeling can complement manual annotation work, reducing cost and time to market while maintaining the quality of massive amounts of data. In addition, automated quality controls can be introduced to measure the quality of the labeled data.

Autonomous vehicles for agricultural applications have grown in recent years as an emerging technology to support productivity and sustainability. Perception sensors integrated with artificial intelligence can be applied to various crop-related applications and support safe automated navigation, environment modeling and crop analysis. This work presents the application of cameras and lidar sensors to classify environmental features applicable to agricultural environments. The intelligent perception framework extracts the structural profile and spectral classification of vegetation that can support various automated tasks that include mowing, harvesting and baling.

In the growing world of connected off-highway machines, new players are offering software solutions/digital services for end customers. Moreover, new business models are emerging, where these new players have to establish a long-term direct relationship with the machine owner/end customer in order to have a profitable business. To address these changes, Bosch Rexroth is setting up a software co-execution platform for connected off-highway machines, where different players can deploy, monitor and share software services. These software services are run on the Linux-based Rexroth Connectivity Unit and remotely managed using the Bosch Device Management Portal.

The operating requirements of modern mobile machines are becoming a challenge even for experienced drivers. Menu navigation, functional complexity and innumerable configuration options make it difficult to get started quickly and to operate the machine in the right situation. Interactive systems to support the driver in adjusting the tractor and implements are already available on the market (e.g. Claas CEMOS) and follow a user-friendly approach; however they lack dynamic change of application (e.g. headland and road travel). A system capable of dynamically identifying not only individual work processes but also corresponding transitions will be presented. The system is based on machine learning principles and feature extractions that are available on the CANbus of mobile machines. AVL will present the process of pattern recognition to identify the actual working process and apply supervised learning based on classified patterns. Furthermore, it will share the experience of applying machine learning on dedicated AI hardware as well as the use of conventional embedded control units to add advanced technology to existing products on the market. An outlook for the development of tractor assistance systems as well as the benefits for OEMs and end customers will also be evaluated.

The Automotive Grade Linux (AGL) community consists of more than 150 companies across the automotive and tech industries who are working together to develop an open-source software platform for in-vehicle applications from infotainment to autonomous driving. The shared software platform provides 70-80% of the starting point for a production project; the remaining 20-30% can be customized to meet the specific needs of industrial vehicles. This presentation will provide an overview of AGL, the roadmap for the future and how to get involved.