DIY robot concept

How to build a tennis robot that catches and returns your shots

Design stage. RoboHub has not built or tested this yet; performance figures are calculations or maker/seller statements.

Difficulty: 5/5 · Expert
Build time: 250–900 hours
Skills: Mechanical fabrication, DC power and fault analysis, Real-time control, Stereo vision calibration
Tools: CAD workstation, Multimeter and current-limited supply, Torque and crimp tools, Radar and reference video

Two experimental catch-and-return designs: a lighter lateral-lane Budget build and a full-size Solid build. Neither has been built or tested.

Generate this build Browse concepts

Safety

Design value: test only under supervision in a separated area after qualified mechanical and electrical review.
Design value: use independent motor interruption, loss-of-link stop and latched manual reset. An app pause cannot replace them.
Design value: a power cut leaves kinetic energy. Verify braking and keep guards locked until independently confirmed standstill.
Design value: use a matched charger under supervision in a cool place. Remove the battery for transport and keep it out of hot vehicles.
Design value: initial testing is limited to 35 °C or cooler in shade. Warmer operation requires component data and thermal tests.

Privacy

Design value: notify everyone recorded, mask neighboring courts and blur faces. Keep clips on the authenticated local network with restricted access and limited retention; no identity recognition.

System diagram

Power

Design: Matched 8S 25.6 V battery in both versions
Design: Main/branch fuses and regulated rails
Design: Independent motor power cut and manual reset

Control

Design: NC stops / wireless loss-of-link / thermal chain
Design: Local compute, app and logs
Design: Bounded real-time drive controller
Design: Court vision host + local link

Motion

Design: Version-matched drive and braking
Design: Slack net, funnel and guarded elevator
Design: Guarded flywheels and NC feeder

Sensing

Design: Triggered court stereo
Design: Robot pose / encoder / IMU
Design: Independent speed/standstill sensing

Design: Matched 8S 25.6 V battery in both versions → Design: Main/branch fuses and regulated rails: Protected DC power
Design: Main/branch fuses and regulated rails → Design: Independent motor power cut and manual reset: Motor branch
Design: Main/branch fuses and regulated rails → Design: Local compute, app and logs: Regulated logic remains on for logs
Design: NC stops / wireless loss-of-link / thermal chain → Design: Independent motor power cut and manual reset: Independent hardwired cutoff; no Linux dependency
Design: Independent motor power cut and manual reset → Design: Version-matched drive and braking: Drive power / brake release
Design: Local compute, app and logs → Design: Bounded real-time drive controller: Timestamped bounded commands with expiry
Design: Bounded real-time drive controller → Design: Version-matched drive and braking: Velocity/torque control
Design: Robot pose / encoder / IMU → Design: Local compute, app and logs: Feedback
Design: Triggered court stereo → Design: Court vision host + local link: Synchronized image streams
Design: Court vision host + local link → Design: Local compute, app and logs: Predicted intercept with uncertainty/expiry
Design: Slack net, funnel and guarded elevator → Design: Guarded flywheels and NC feeder: Same captured ball
Design: Independent motor power cut and manual reset → Design: Guarded flywheels and NC feeder: Motor power
Design: Independent speed/standstill sensing → Design: NC stops / wireless loss-of-link / thermal chain: Overspeed/fault/zero interlock
Design: Bounded real-time drive controller → Design: Guarded flywheels and NC feeder: Calibrated launch; base must be stationary

The product goal is physical catching and return of the same ball.

The product goal is physical catching and return of the same ball. A fixed launcher, repositioning and reserve-ball virtual catches are development stages, not the finished rally robot. All proposed architecture and performance targets here are design values.

Budget uses a 2.4 × 1.5 m slack net, a lateral-only base and a 20-ball reserve for constrained slow rallies.

Budget uses a 2.4 × 1.5 m slack net, a lateral-only base and a 20-ball reserve for constrained slow rallies. Design values: 23.611 kg allocated mass, 60 km/h nominal incoming shots and 250–600 build hours. At 18 m and 60 km/h, stopped-to-stopped travel calculates to about 0.344 m before net coverage; this is not full-court coverage.

Solid uses external court-side stereo vision, a larger drive system and guarded ball recovery.

Solid uses external court-side stereo vision, a larger drive system and guarded ball recovery. Design values: 59.786 kg allocated mass and 400–900 build hours. Catch rate, assembled weight, runtime, thermal behavior and transport fit still need physical tests.

Maker-stated: Acemate S10 and S10 Pro catch and return shots.

Maker-stated: Acemate S10 and S10 Pro catch and return shots. Check their catalog pages for dated price evidence. Neither incomplete DIY subtotal demonstrates a cheaper or equivalent robot.

Original and build

Independent comparison; no affiliation with or endorsement by the maker is implied.

Original robot	This build
Main functionMaker-stated: Acemate S10 catch-and-return; S10 Pro is another complete option.	Design value: physical capture and return required; no measured catch rate.
MassMaker-stated S10: 17.8 kg.	Design value: Budget 23.611 kg; Solid 59.786 kg. Neither is weighed.
MovementMaker-stated S10: up to 5 m/s.	Design value: Budget 1.8 m/s lateral only; Solid 3 m/s target, subject to braking tests.
Launch speedMaker-stated S10: up to approximately 97 km/h.	Design value: Budget 30–70 km/h; Solid 30–120 km/h after calibration and rotor limits.
RuntimeMaker-stated S10: up to 3 h, varies by mode.	Design calculation: Budget 35–52 min at 200–300 W; Solid 46–84 min at 250–450 W. No runtime measured.
Reserve capacityMaker-stated S10: 80 balls.	Design value: Budget 20 balls; reserve feeds never conceal failed physical catches.
Price contextMaker-stated price evidence is linked from the Acemate S10 catalog page.	Dated priced-parts subtotals are incomplete; missing fabrication, required items and delivery prevent a complete price comparison.

Prefer a ready-made robot?

Acemate S10

Maker-stated catch-and-return; compare a complete robot before committing to development.

Acemate S10 Pro

Maker-stated catch-and-return in the Pro version; compare its separate specifications.

Tenniix Tennis machine (Basic / Pro)

Compare the maker’s Basic and Pro launcher variants.

Tenniix S tennis machine (S Basic / S Pro)

Compare the maker’s S Basic and S Pro launcher variants.

Tennibot Partner V2 (tennis)

Compare the ready-made tennis launcher configuration.

Tennibot Rover

A ball collector; it is not a catch-and-return opponent.

Live AliExpress prices for the BOM

Pulling current matching products on AliExpress…

Sources and validation

Product claims, safety boundaries and the capture architecture are linked to their original sources. Marketplace prices remain live and may change.

Acemate S10 maker specificationsMaker-stated specifications and list-price snapshot, 10 November 2026.Acemate S10 Pro maker specificationsMaker-stated; separate product and operating-mode limits.Raspberry Pi 5 product documentationMaker-stated compute requirements; integration remains a design value.

Design variants

FAQ

Has RoboHub built and tested this?

No. This is a design-stage plan. Acceptance targets, estimated hours and calculations are design values, not measured results.

Is the parts subtotal the complete build cost?

No. Dated candidate prices leave required unpriced parts, workshop work, labor, tax and delivery unresolved. Optional items are excluded from the required-parts subtotal.

Can the app replace an emergency stop?

No. Design value: independent hardwired motor interruption, loss-of-link handling and manual reset must work without the app or Linux.

Does Budget cover a whole singles court?

No such performance is established. Design value: use a constrained lateral lane, recenter between shots and reject targets beyond measured timing and uncertainty limits.

How should catch rate be tested?

Design value: test 100 shots per speed, distance and offset cell. Count aborted and unreachable balls in the all-shots denominator; report virtual catches separately and publish uncertainty.

Why use cameras beside the court?

Design value: a calibrated 4 m baseline can reduce depth uncertainty relative to a short onboard stereo baseline. Effective frame rate, exposure, synchronization and processing latency must all be measured.

Related robot guides

Turn this concept into a sourced build

Start with this prompt prefilled, then let RoboHub generate the live parts list, wiring plan, CAD and firmware.

Generate build

How to build a tennis robot that catches and returns your shots

Power

Control

Motion

Sensing

Versions

Parts

Budget

Solid

Steps

Budget

Define the rally lane

Build and weigh the modules

Test power and the independent stop chain

Measure traction and braking

Calibrate the court-side cameras

Catch and return the same ball

Publish measured acceptance results