A New Perspective on the Levels of Autonomy

As the conversation around autonomy and its capabilities continues to evolve, the importance of distinguishing its applications between on-road and off-road vehicles becomes increasingly urgent. Consumer interest in autonomous functionality and improving on-road safety both in and out of the vehicle is a different motivator than labor challenges, terrain considerations, rising fuel costs, etc. that face off-road vehicle operators. Autonomy for on-road vehicles has the primary goal of driverless passenger vehicles. But autonomy for off-road vehicles or equipment encompasses broader functionality, e.g. compacting soil or spraying crops. When we talk about autonomy, the challenge resides in the various definitions, usage and meaning of the word “autonomy.” Questions abound: 

• Is automation the same as autonomy? 
• Is it driver assistance? 
• Is precision blade control on a dozer autonomy? 
• What about a connected worksite or command-and-control system? 

At Trimble, the short answer is “yes.” We see autonomy as all of these varying points on the journey to more intelligent, and more efficient, workflows.

The widely accepted view of the 5 levels of autonomy, created by the Society of Automotive Engineers, were in reference to autonomous passenger cars from the perspective of the driver and/or operator. These 5 levels are defined as:  

• Level 1: driver assistance, everything on
• Level 2: partial automation, hands off
• Level 3: conditional automation, eyes off
• Level 4: high automation, mind off
• Level 5: full automation, steering wheel optional 

This framework works well when the objective is to get a vehicle from point A to point B. For Trimble customers, navigating and steering a vehicle is only a small part of their workflow, and is rarely the end-goal. 

As a technology partner across a range of industries, Trimble also measures the success of autonomous applications by the volume of work done, be it the amount of earth moved, the number of acres fertilized and by the number of complicated tasks occurring simultaneously that the operator must perform or oversee. While the 5 levels still applies, for the off-road space it may be better to organize our thoughts around a more encompassing series of stages:

1. Operator assistance - The operator or driver is reliant on a light bar or visual indicator to provide information in realtime of where the vehicle or implement should be. This guidance will inform decision making and improve efficiency. Examples include collision warning, or notification of an object within the path.
2. Task Automation - A smart system is capable of automatically completing a part of the job. Current task automation examples include blade control on a dozer, variable spraying rate on a tractor and lane keeping across a range of vehicles. 
3. Supervised Autonomy - This is a higher level of automation where the operator is in a supervisory role but is still responsible for reacting to unexpected conditions. Both the Horsch sprayer and Dynapac soil rollers demonstrate this level of autonomy.
4. Full Workflow Automation - The end goal of fully autonomous capability and full workflow automation includes intelligent autonomy, optimizing for a business need and/or operating without direct supervision on site. Full autonomy can encompass a vehicle or machine that is given a task and is responsible for efficient and safe completion, as well as extending to full site autonomy, e.g. a connected worksite, TC1 or connected farm.

Trimble looks at autonomy not as a stack of technologies, but as a matrix of capabilities that may fit into more digestible groups like “perception” and “control systems.” Within these groups, there are many separate and distinct building blocks. Detecting an object on a machine’s path is not the same as classifying a scene for path planning. The importance of each building block capability, or node in the matrix, varies according to the stage as defined above, the complexity of the task and the environment itself.

What is the significance of Trimble’s approach to autonomy? We recognize not all applications of autonomy need the same capabilities at the same pace or with the same scale. As Trimble continues to develop new solutions, encompassed within our market-leading platform, and expand our market reach, we are able to offer our clients and partners a range of solutions and capabilities to meet them wherever they may be on their journey to autonomy. Trimble brings experience and expertise to guide our partners toward their autonomous goals.