Development Progress
Follow along as we build the most compact, durable, and intelligent tennis ball machine ever. I'll be posting regular updates about our development process, challenges, and milestones.
May Update: Draw Winner, Development Progress, and My Take on the New Kickstarter Campaigns (AceMate, Tenniix)
May Draw Winner Announcement
Congratulations to Delice for winning our May $100 Tennis Warehouse gift card draw!
Drawing results from May 9, 2025
$100 Tennis Warehouse Gift Card
Delice, I will reach out to you in a separate email and have this gift card sent to your email.
Thank you to everyone who participated in our survey and referred friends! Your feedback is invaluable as we continue to develop our tennis ball machine.
Sorry about the delay in this month's draw. We've been working hard on getting the mechanical parts ready and developing the app. Next month we'll be on time on June 1st. So don't wait, complete the survey today and refer friends to increase your chances of winning our next $100 Tennis Warehouse gift card!
Development Progress
Hardware
We're making exciting progress on the hardware front!
We've successfully assembled our first prototype using RoboMaster motors and 3D printed parts to test our "dual flywheel launching" concept. The results are promising - we achieved launch speeds of ~75mph, confirming that our design direction is on the right track. For the pitch and oscillation mechanisms, we implemented custom-designed and 3D-printed gears, though we did identify some durability challenges that need to be addressed.
Based on these learnings, we're now advancing to the next development phase. Instead of continuing with off-the-shelf motors (like the RM 3508), we've designed custom motors and sent the specifications to a specialized manufacturer. On the mechanical side, one key improvement is the implementation of a worm gear shaft system, which should significantly enhance durability and control precision. Additionally, we're testing a new flywheel design that promises to further reduce the overall volume and weight of the machine - making it even more portable and user-friendly.
First prototype design with RoboMaster motors and custom gears
Close-up of the dual flywheel launching system
Software
On the software front, we've made significant progress! We've developed a functional app that can detect player position and poses in real-time, and through our custom reasoning module, determine the optimal ball placement for your practice session.
We've also started experimenting with training custom models to better adapt to various lighting conditions and camera positions, which will ensure consistent performance whether you're playing indoors, outdoors, or in changing weather conditions. If you have expertise in computer vision, I'd love to connect and discuss potential collaborations!
Now that the core functionalities of phase 1 are developed and being tested, our next focus is enhancing the user interface and experience. If you're interested in being an early tester, please reach out and I can add you to our TestFlight (sorry it has minimal UI for now).
I've been monitoring feedback from users of other ball machines, and features like offline editing mode and real-time ball speed and spin data displaying on app are often requested. Don't worry, these will all be standard features in our app - we are the only team on the market that understands both technology and tennis.
Other Updates
We expect to have several EVT (Engineering Verification Test) units ready by late July or early August. Once these are shipped to the States, we plan to host several user testing events in the Bay Area this summer. If you're local and interested in being among the first to try our prototype, please email us at hello@rallie.tennis. We're particularly looking for players across different skill levels (from beginners to advanced) to provide diverse feedback.
My Thoughts on Acemate and Tenniix
As some of you might have noticed, there are two projects that recently launched Kickstarter campaigns. These are:
Both feature autonomous mobility (with perception, computing, and Mecanum wheels) for "rallying style" playing. This isn't really an innovative concept - it's actually what everyone in the tennis robot field is ultimately trying to achieve. However, I believe we're not there yet. Robot design is a complex system design challenge with some crucial design choices to make.
To achieve real rallying capability as advertised, you will need:
- Perception: multiple cameras combined with other sensors (to account for bad weather conditions when cameras can't see clearly)
- Processing & Planning: strong processing unit (AM is using RK3588 which might not be enough for applications that require low latency - athlete level reaction time)
- Execution: strong and agile wheels that move omni-directionally at 5m/s and with great suspension and control
- These need to be considered within physical limitations such as weight and battery (these two are conflicting themselves) and cost. Finding a balance of all these is the art of hardware design and exactly why I like hardware so much :).
From what I'm seeing in the pictures, both products present significant risks:
- Durability of the wheels - Mecanum wheels aren't designed for outdoor tennis court surfaces. At speeds of 5m/s, these wheels would experience significant stress and wear, especially on clay or hard courts. The small rollers that enable omnidirectional movement are particularly vulnerable to debris and uneven surfaces. Additionally, using these wheels for oscillation adds further mechanical stress to an already compromised design.
- Battery limitations - Current ball machines (Lobster, SpinShot etc.) have around 4 hours battery life, using their bulky premium batteries. Imagine powering all these additional units: 2 4K cameras, likely other sensors such as ultrasonic or single-thread lidar, 4 additional motors each moves one wheel, and a processing unit that's always running at top capacity. The batteries in the picture look significantly smaller than Lobster battery, and AM is claiming to have the same battery life - you can make your judgment now.
- Overadvertising - Both seem to be showy marketing instead of real production-level products. The marketing has raised user expectations and users are expecting real rallying robots that return most balls, just like those in sci-fi. Clearly managing expectations would be better.
Even ChatGPT agrees that Mecanum wheels at 5m/s (18 km/h) is problematic, especially for tennis applications
As someone who's been trying to find balance between portability/weight, battery life, computing power, and user experience for my entire career, and someone who worked on fully autonomous field robotics in outdoor harsh environments for the past year, I don't see a real rallying robot (that catches at least 80% of incoming balls and returns balls like a real human would) that's durable (think no repair and maintenance for at least 2 years) on outdoor courts with compact design (<20kg) and 4+ hours of battery life, and at reasonable prices, coming in the next 2 years.
This is exactly why I chose not to go with the fancy self-driving robots for my first product. Let's wait and see what the real user reviews on these look like, and learn from these products, before we start building something in the likings of these.
That's all the updates for this month. I will be back with possibly some videos of our tests next month, and maybe an ID concept design for you guys to comment on. Until then, take care!
Development Update 3/22 - 3D printed parts, motors and other issues
The 3D printed parts have arrived! (maybe it's time to buy a 3D printer. Anyone has suggestions?) There's some issue with the precision of these components, particularly the section for ball ejection. We'll need to reprint these parts to ensure a proper fit and function.
Additionally, we've successfully installed the yaw and pitch to change height and get oscillation. However, since we used some second-hand motors (on a tight budget for the prototype), these motors exhibit significant backlash, affecting their control capabilities. The controller works fine though. Will need to buy some new motors.
While we await the arrival of the newly printed parts, we will proceed with soldering the necessary wiring. We also plan to replace the problematic motor to ensure the machine operates smoothly and meets our standards for quality and performance.
I'll get back soon with updates and hopefully some videos to show! Keep updated!
Project Kickoff: The Journey Begins
Today marks the official kickoff of the rallie tennis ball machine project! After months of research, planning, and initial designs, we're finally ready to begin the development process in earnest.
I've finalized the core specifications for our first prototype and placed orders for the custom components we'll need. Our goal is to create a machine that's not just another ball launcher, but a true training partner for tennis players of all levels.
I've also set up this progress page where I'll be sharing regular updates about our development journey. From technical challenges to exciting breakthroughs, you'll get an inside look at what it takes to bring a new product to life.
Help Shape Our Product
Your feedback is invaluable as we develop our tennis ball machine. Take our survey to share your experiences and preferences - and you could win a $100 Tennis Warehouse gift card!
Take Survey and Win $100Join Our Waitlist
Be the first to know when our revolutionary tennis ball machine launches. Enter your email below to join our waitlist and receive exclusive updates.