The Julius project

Building autonomous kayaks for science and exploration.

Similar projects worth following
The Julius project is an attempt to determine whether "end-of-life" white water kayaks can provide a suitable, low cost, platform on which to build an autonomous vehicle capable of undertaking scientific research and exploration missions.

Autonomous boats are still either expensive, or cheap but patched together hobby projects. We want to provide the building blocks to enable any organisation or person to take the electronics and software, find a suitable body and be up and running as easily as possible.

Why Kayaks?
Well, for one thing - there are a lot of cheap used ones around. When you don't care about the seats and pads, and only whether it floats then you can grab a bargain.

Second - there are a lot of styles targeted at specific types of water, so you can pick or switch to the most suitable for your needs at the time.


For an upcoming expedition we need to survey relatively shallow waters and coastline. Parts of the coast are too shallow to use even a small ROV like the OpenROV or even to snorkel - so we needed a vehicle with a shallow draft that would also be stable enough to handle small waves.

Existing technology

There are, admittedly, quite a few autonomous marine vehicles around at the moment. A lot of them are very expensive and out of reach of small organisations and individuals. Some are a lot cheaper, but are aimed at ocean crossings or long journeys where large waves and the threat of being overturned mean that a deep draft or keel are required.

We needed something with a shallow draft that was capable of hugging the coast.

Why Kayaks?

There are a lot of Kayaks available specifically designed to handle different types of water - from flat bottomed white-water kayaks that are capable of turning sharply, through to longer sea kayaks more suited to travelling in long straight lines quickly.

Because of the popularity of kayaking as a hobby, there are a lot of older, end of life, kayaks available to purchase cheaply on places like Facebook and eBay.

As we aren't concerned with the condition of the seats or outfitting, then we can pick up perfectly capable Kayaks that others might not consider.

Why Raspberry Pi?

The Raspberry Pi is an extremely low cost, yet very capable computer - the Raspberry Pi Zero even more so. A lot of people already own at least one, and schools tend to own a few for educational use.

It made sense to use something that a lot of people would be familiar with. Whilst there are existing autopilot boards available, some of them open source, for a lot of sensor recording and reading - a Pi or similar would be needed anyway.

Modular approach

Not all waterways are the same, and not all survey missions have the same conditions. Sometimes you may need something that can keep a straight line easily, other times something that can turn quickly to get around a rocky coastline.

By keeping all of the electronics in a self-contained unit, we hope to be able to simply swap the body of Julius to best suit the surveys that need to be done.

Placing him in a 13-15 foot long sea kayak with a skeg will help with long straight line surveys, but remove the ability to make tight turns.

Wheras placing him in a 6 foot white water kayak means that he will be able to turn very quickly to follow a winding or rocky coastline, but not have as much straight line speed or be able to handle larger waves.

Open source

To the best of our ability, everything we use, write or build will be open sourced. By the end of this project we aim to have a detailed Bill of Materials, designs for new hardware and assembly instructions so that anyone interested will be able to build, and hopefully improve on, the Julius system.

  • 1 × Old White Water Kayak Condition of the outfitting / internals isn't important - as long as it floats and has few (or better yet, no) dents, deep cuts or welds.
  • 2 × Propellers We're using BlueRobotics T100 properllers
  • 1 × Raspberry Pi To Keep things cheap, we're using a Raspberry Pi for processing and control
  • 1 × Motor Controller We developed our own controller for ESC powered motors call the ESCAPE board

  • Promotional Video

    Dark Water07/10/2017 at 17:02 0 comments

  • Ready for testing

    Dark Water07/03/2017 at 15:17 0 comments

    ... and then it was off to test - more on this soon.

  • Electronics box tidy up

    Dark Water07/03/2017 at 15:14 0 comments

    The electronics still look a bit of a mess, but they are tidier than at the start. We'll be concentrating and getting the internals for this a lot neater.

    Inside you can see the Raspberry Pi 3, two ESCs for the motors, an RC receiver for manual control and mode switching and GPS connected to an external antenna for location.

  • Testing the Box fit

    Dark Water07/03/2017 at 15:11 0 comments

    Once the structure is assembled and sturdy the boxes will slot in.

    We're currently placing the battery box near the rear, but might switch the two in future as it's easier to access the rear box and that would be handy for getting at the electronics.

  • Building the Internal support

    Dark Water07/03/2017 at 15:09 0 comments

      The internal support structure inside the Kayak serves two functions:

      1. To hold the battery and electronics boxes in place
      2. To create a space below the battery and electronics boxes that can be used for cabling and other sensors that need to be attached to the base of the Kayak

      We made the support structure out of aluminium extrusions attached together with some brackets created from 5mm HDPE sheets.

      Once we had the spacing correct it was placed inside the Kayak and bolted to cross supports attached to the screws that were previously used to hold the seat in place.

      We then created small, adjustable, support feet from bolts and door stoppers so we could ensure that the internal structure was level.

  • We've been building and testing...

    Dark Water06/26/2017 at 15:06 0 comments

    We've been doing a lot of work on Julius, but haven't had the chance to update the project log - we've got a bit of video to edit, and a lot of writing to do after a recent testing trip to Ireland - but in the meantime, here is a little video clip.

  • A little bit of processing

    Dark Water10/03/2016 at 19:50 0 comments

    We've been working on the video footage taken from the Kayak over the Manchester weekend - our first pass dense point cloud of lock 89 is below.

  • Manchester Waterside Festival

    Dark Water09/24/2016 at 19:52 0 comments

    I've still got a lot of video and images to process, but I've started to put together some clips from the Manchester Waterside Festival.

    We ran Julius without a tether for the weekend and he performed very well - we managed to get some above water footage with the GoPro camera and quite a bit of underwater footage with the 360 degree camera. Though as we were operating in a canal, it was a bit murky - I'll have to see what can be pulled out of it.



  • See Julius in Manchester

    Dark Water09/08/2016 at 10:24 0 comments

    We'll be in Manchester (the UK one) this weekend (10th and 11th September 2016) and we'll be taking Julius with us, so if you are in the area then pop along and say hello.

    There are free events along the canal all weekend - we'll even show you how to build and pilot a Lego underwater robot.

    More details are here:

  • Neater cabling and more

    Dark Water09/06/2016 at 21:51 2 comments

    A lot of work has taken place since the last update.

    We now have the motor wiring going through the bottom of the hull using some Blue Robotics hull connectors - this makes things a lot neater and removes the motor wiring that was wrapped around the outside of the Kayak.

    We've also, momentarily, switched to using a Raspberry Pi 3 so we can take advantage of the on-board WiFi. We may yet switch back to a Pi Zero, we'll be running some tests to determine suitability.

    The batteries ( two 3S LiPo battery packs ) are currently in the same box as the electronics, we're going to switch them to their own box in the near future.

    We're now running a custom build of the Ardupilot software ( the Ardurover version ) and have everything communicating with the ground control software over WiFi. The initial test run of the new software went well, though we need more manual control practice.

    We're waiting for a connector for our external GPS antenna to arrive and then it'll be time to head to a larger body of water for autonomous testing.

View all 18 project logs

Enjoy this project?



Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates