Case study

Thoro.ai

Thoro.ai

Managing fleets of autonomous robots in large-scale environments requires more than just hardware; it demands a seamless, real-time orchestration layer. Thoro.ai stepped in with a vision to make robot management intuitive: what if operators could monitor and control every unit in a warehouse with the same ease as a single device?


The platform needed a development partner who could:

  • Build a high-performance orchestration engine for real-time fleet control
  • Enable low-latency communication between the cloud and mobile robots
  • Create a rich, interactive visualization of warehouse operations
  • Implement a highly scalable and secure recursive permissions system
Industry
Robotics & Automation
Country
USA
Services
Web Development, Software Architecture, IoT Integration
How we leveraged Elixir and Phoenix LiveView to build a high-performance orchestration system for real-time robotics monitoring and control.

Technologies

Elixir
Phoenix LiveView
Postgres
Terraform
AWS
C++

Project Overview

Thoro.ai is a leader in autonomous navigation software, turning manual industrial equipment into self-driving robots. To support their growing global presence, they needed a robust Fleet Management Platform. Curiosum was tasked with building the "brain" of this operation - a system capable of managing hundreds of autonomous devices simultaneously, providing operators with instantaneous control and a comprehensive overview of their warehouse environments.

The Challenge

Managing hardware via the cloud presents unique challenges: high latency, connection stability, and the need for immediate feedback. Thoro.ai required a platform where users could summon robots, monitor statuses, and manage complex organizational hierarchies without the lag typically associated with web-based hardware management.

No items found.

The Solution: The Power of Elixir & LiveView

Thoro had previously been using a platform heavily relying on a serverless infrastructure, which was no longer scalable enough for the product’s needs. For the new system, Elixir was specifically picked for its unmatched ability to handle concurrent processes and Phoenix LiveView to deliver a desktop-like user experience directly in the browser. This stack allowed us to maintain persistent, low-latency connections between the robots and the management dashboard.

Building the system’s features, we’ve used Elixir and the BEAM’s comprehensive toolset (GenServers, Broadway, Phoenix LiveView, libcluster, and many others) to create a system that feels responsive at the operator’s and manager’s end, is efficient, scalable and reliable at the backend, and secure and flexible in its access control procedures across the globe.

Warehouse visualization & real-time robot summoning

Using the Phoenix LiveView stack, we built a warehouse visualization dashboard that allows operators to see the robot fleet in real time, monitor their statuses, and summon robots as needed. The Control Tower module displays robots' positions in realtime overlaid on the actual site map retrieved from the device's positioning system and allows oversight of each machine's functioning, including camera view if provided by the robot, as well as debugging behavior issues using telemetry data exchanged via MQTT. Using the Callbox feature, operators can summon robots to waypoints defined on the map and review relevant statistics about the given zone and waypoint.

Workflow engine

We built a custom workflow engine that lets operators define, reuse, and execute complex multi-step robot missions without writing a line of code. The engine handles the full coordination loop — from dispatching a robot to a pickup zone using the VDA5050 protocol through to confirming delivery and returning it to queue — while keeping the cloud and the robot each responsible for exactly the right layer of the operation. The result is a system that's both flexible enough to adapt to changing warehouse conditions in real time and robust enough to scale across an entire robot fleet without operator bottlenecks.

Permission management system

The platform includes a system that allows fleet administrators to manage cascading permission hierarchies across the robot fleet, with fine-grained control over who can do what at which level of the organization. The custom solution we built helps Thoro support, device and vehicle vendors, tenants, and operators define access control rules in a variety of ownership scenarios and organizational structures.

The Result

The partnership between Curiosum and Thoro.ai resulted in a high-concurrency platform that stands out in the robotics industry. By choosing Elixir, we ensured that the platform is not only fault-tolerant but also highly scalable, ready to support the next generation of autonomous warehouse solutions.

Client’s voice

Curiosum has been indispenable to Thoro in growing our cloud and remote-management features for our expanding worldwide fleet. Their extremely capable engineers have fit seamlessly into our team, and have delivered significant value for us and our customers. I would highly recommend them to anyone.
Thoro.ai
Joseph Hosteny
Head of Cloud and Fleet Systems
,
Thoro.ai

Let’s build something together

Ready to turn your curiosity into a successful digital product?
Share your idea, and let's explore the possibilities.

Our experts worked with 20+ worldwide tech innovators

Szymon Soppa
CEO