←back to thread

Launch HN: Flywheel (YC S25) – Waymo for Excavators

103 points jashmota | 3 comments | 24 Sep 25 16:48 UTC | HN request time: 0.64s | source

Hey HN, We're Jash and Mahimana, cofounders of Flywheel AI (https://useflywheel.ai). We’re building a remote teleop and autonomous stack for excavators.

Here's a video: https://www.youtube.com/watch?v=zCNmNm3lQGk.

Interfacing with existing excavators for enabling remote teleop (or autonomy) is hard. Unlike cars which use drive-by-wire technology, most of the millions of excavators are fully hydraulic machines. The joysticks are connected to a pilot hydraulic circuit, which proportionally moves the cylinders in the main hydraulic circuit which ultimately moves the excavator joints. This means excavators mostly do not have an electronic component to control the joints. We solve this by mechanically actuating the joysticks and pedals inside the excavators.

We do this with retrofits which work on any excavator model/make, enabling us to augment existing machines. By enabling remote teleoperation, we are able to increase site safety, productivity and also cost efficiency.

Teleoperation by the operators enables us to prepare training data for autonomy. In robotics, training data comprises observation and action. While images and videos are abundant on the internet, egocentric (PoV) observation and action data is extremely scarce, and it is this scarcity that is holding back scaling robot learning policies.

Flywheel solves this by preparing the training data coming from our remote teleop-enabled excavators which we have already deployed. And we do this with very minimal hardware setup and resources.

During our time in YC, we did 25-30 iterations of sensor stack and placement permutations/combinations, and model hyperparams variations. We called this “evolution of the physical form of our retrofit”. Eventually, we landed on our current evolution and have successfully been able to train some levels of autonomy with only a few hours of training data.

The big takeaway was how much more important data is than optimizing hyperparams of the model. So today, we’re open sourcing 100hrs of excavator dataset that we collected using Flywheel systems on real construction sites. This is in partnership with Frodobots.ai.

Dataset: https://huggingface.co/datasets/FlywheelAI/excavator-dataset

Machine/retrofit details:

  Volvo EC380 (38 ton excavator)
  4xcamera (25fps)
  25 hz expert operator’s action data
The dataset contains observation data from 4 cameras and operator's expert action data which can be used to train imitation learning models to run an excavator autonomously for the workflows in those demonstrations, like digging and dumping. We were able to train a small autonomy model for bucket pick and place on Kubota U17 from just 6-7 hours of data collected during YC.

We’re just getting started. We have good amounts of variations in daylight, weather, tasks, and would be adding more hours of data and also converting to lerobot format soon. We’re doing this so people like you and me can try out training models on real world data which is very, very hard to get.

So please checkout the dataset here and feel free to download and use however you like. We would love for people to do things with it! I’ll be around in the thread and look forward to comments and feedback from the community!

Show context
seabrookmx ◴[24 Sep 25 19:04 UTC] No.45364639[source]▶
> The joysticks are connected to a pilot hydraulic circuit, which proportionally moves the cylinders in the main hydraulic circuit which ultimately moves the excavator joints

I've actually spent a decent amount of time running an excavator, as my Dad owns a construction / road building company. It was a great summer job!

An important note about the pilot hydraulics is that they _provide feedback to the operator_. I would encourage any system that moves these controls on behalf of a remote human operator or AI to add strain gauges or some other way to measure this force feedback so that this data isn't lost.

The handful of "drive by wire" pieces of equipment that my Dad or other skilled operators in my family have ran were universally panned, because the operators are isolated from this feedback and have a harder time telling when the machine is struggling or when their inputs are not sufficiently smooth. In the automotive world, skilled drivers have similar complaints about fully electronic steering or braking systems, as opposed to traditional vacuum or hydraulic boosting approaches where your foot still has a direct hydraulic connection to the brake pads.

replies(4): >>45365403 #>>45365849 #>>45366303 #>>45366966 #
1. jashmota ◴[24 Sep 25 21:40 UTC] No.45366303[source]▶
You're right! This is exactly why we like to do mechanical actuation - we are able to achieve bilateral telepresence, which essentially gives the torque (haptic) feedback over the internet! So on small excavators, you can absolutely feel the resistance. We also stream the engine audio, which tells you how hard the hydraulic pump is working. Operators like our system for these reasons :)

I'd like to get a chance to talk to you and your Dad to get feedback. How do I reach you? My email is contact at useflywheel dot ai

replies(1): >>45372030 #
2. IgorPartola ◴[25 Sep 25 12:33 UTC] No.45372030[source]▶
Not my industry at all.

I am curious if something like this is an opportunity for a whole new type of controls and feedback. Since the operator doesn’t have to be in the excavator physically they could take on any position: standing, sitting, lying down, etc. Instead of sending haptic feedback to the joystick it could be sent to a vibrating wrist band. You could hook up the equivalent of a Nintendo Power Glove to have the scoop operated by the operator simulating scooping action. Turning the excavator can be controlled by the operator turning their head and moving it around can be done by the operator walking on an infinite treadmill. Motor strain can be done via color of light or temperature rather than sound. You could have a VR helmet that can also show you a birds eye view from a companion drone, overlay blueprints, show power and water lines, measure depth, etc. I don’t know if it is possible but maybe you could even measure soil composition somehow to show large rocks, soft soil that is dangerous to drive over, inclination angles where the excavator is about to drive, etc.

I imagine skilled operators prefer familiar controls but perhaps there are interesting improvements unlocked by exploring alternatives. It might also fundamentally change how accessible it is for non-professionals to use these machines. I rented an excavator from Home Depot a few years ago to dig a foundation and the learning curve was not shallow. I wonder if a more “natural” interface would help keep people safer.

replies(1): >>45381020 #
3. jashmota ◴[26 Sep 25 00:15 UTC] No.45381020[source]▶
These are really really interesting thoughts around the remote teleop interface. A few things you mentioned have been on my mind. Teleop UX is underdeveloped and its affects is underestimated, and it could turn out to be a huge thing for us and humanoid companies if autonomy is harder than it seems now. I don't believe construction equipment operation today is optimal. We just go with it as that's what we have. It's far from intuitive and easy for a newbie to start operating to give some context and requires hours of practice. There would be a bit of training to be done for remote teleop as well. Might as well make it easy to interface and improve on the existing experience, which hasn't really evolved for years and decades!

I'd like to have a chat with you if you're up for it: contact at useflywheel dot ai