Episode #2
Digital Ag Global, season 3, episode 2, Building autonomy in agriculture
What does it really take to move agricultural autonomy from an impressive demonstration to a reliable, commercially viable system working in real fields? Join the next episode of the Digital Ag Global podcast, “What It Really Takes to Build Autonomy in Agriculture,” broadcast live on LinkedIn and YouTube on September 23 at 9:00 AM ET. Our guest is Michael McGuire, CTO of EarthSense, where he leads the development of AI perception and autonomy systems for agricultural robots and machinery. Michael has worked on autonomous robotic fleets, tractors, and commercial oil palm fertilization systems across the United States and Southeast Asia. His experience covers the entire journey from prototype to production: computer vision, localization, navigation, systems architecture, hardware integration, manufacturability, field reliability, and the operational workflows required to make autonomous technology practical for farmers. In this episode, we will explore what “real autonomy” means in agriculture—and why a robot completing one successful demonstration is very different from a machine that can work reliably through heat, dust, mud, dense crop canopies, unreliable connectivity, and changing field conditions. The conversation will cover: • What is inside an agricultural autonomy stack, from perception and localization to navigation, safety, edge computing, and data • How vision-based navigation can operate beneath dense crop canopies without relying entirely on RTK GPS • The roles of cameras, LiDAR, radar, sensor fusion, and low-cost perception systems • Why traversability, terrain, mud, slopes, vegetation, people, and animals make off-road autonomy uniquely difficult • How autonomous systems should degrade gracefully and stop safely when conditions change • What it takes to move from a field prototype to a reliable commercial product • Why hardware durability, manufacturing, maintenance, field service, and unit economics can matter as much as algorithms • Where autonomy is already creating measurable value in fertilizing, spraying, orchards, vineyards, and oil palm plantations • How per-plant precision can improve input application, coverage, traceability, and operational visibility • Why service-based and retrofit models may accelerate adoption • Which autonomous agricultural use cases are commercially realistic today—and which remain aspirational We will also discuss the “autonomy ladder”: beginning with better field records, moving into driver assistance, and eventually reaching full autonomy where the economics and operating conditions justify it. Michael will share lessons from deploying agricultural robots in demanding environments, including the transition from research robots in Illinois to commercial systems operating on palm estates in Southeast Asia.
