Source-linked AI summary

From Sky to Soil: A Morphing Aerial-Ground Robot for Seed Deployment

Namai Chandra, Lining Yao

arXiv:2609.37072v1cs.RO

TL;DR

Surface seed broadcasting offers limited control over seed placement in soil, despite aerial vehicles’ access to remote restoration sites. This paper presents a morphing aerial-ground robot that reuses its propulsion motors for drilling. In outdoor trials, seeds entered the borehole in 47 of 50 cycles, while the prototype also demonstrated ground repositioning.

  • Problem

    Aerial seed delivery reaches remote sites but does not actively control whether seeds are covered by soil.

    Source: §I · p. 1
  • Method

    A morphing robot folds its rear arms into ground wheels and lowers a tray to couple propulsion motors to a drill for seed embedding.

    Source: §I · p. 2
  • Results

    47/50 outdoor drill-and-dispense cycles placed a seed in the borehole; germination was not evaluated.

    Source: E. Seed Placement Trials · p. 7
  • Takeaways & Limitations

    The hardware prototype demonstrates flight, ground repositioning, and controlled-depth seed embedding in a single aerial-ground platform.

    Source: §IX · p. 8

    No controlled wind or disturbance testing was performed, and tray descent and drill engagement assume a roughly level landing attitude.

    Source: §VIII · p. 7

Abstract

from arXiv · show

Aerial seed broadcasting can reach remote restoration sites, but provides limited control over seed placement within the soil. This paper presents a geometry-assisted, tri-functional morphing robot that combines aerial access, ground locomotion, and controlled-depth seed embedding in a fly-drive-plant architecture. After landing, the platform reconfigures into a four-wheeled planting configuration: an electronically coupled dual-motor drive folds the rear arms outward to form ground wheels, while a descending front tray engages the propulsion motors with a drill gear train. Reusing the propulsion motors for drilling eliminates a dedicated drill drive. The planting sequence forms a hole, dispenses a seed, and allows the vehicle to reposition on the ground or return to its flight configuration. Ground mobility supports repeated planting without requiring a separate flight between adjacent sites. A companion controller issues reconfiguration, tray, and seed-gate commands, while a dedicated autopilot handles flight control. The morphing and planting mechanisms are validated using a hardware prototype that demonstrates ground repositioning and seed embedding. This proof of concept establishes a hardware basis for aerial-ground seed embedding through coordinated reconfiguration and actuator reuse.

I. INTRODUCTION

The paper addresses the gap between aerial access to remote restoration sites and precise seed placement within soil. It presents a fly-drive-plant prototype and situates it among aerial, hybrid, and seed-delivery systems that do not demonstrate controlled-depth seed embedding.

I. INTRODUCTION

  • Surface broadcasting can leave seeds exposed; one field study reported 1.4%–3.3% establishment after surface broadcasting versus 22.6%–33.4% after shallow burial.The paper identifies desiccation and predation as likely causes of poor surface establishment.Source: §I · p. 1
  • The prototype combines aerial access, ground repositioning, and controlled-depth seed embedding through a morphing architecture that reuses propulsion motors for drilling.Rear arms become ground wheels, while a descending front tray couples the propulsion motors to the drill gear train.Source: §I · pp. 1–2
  • The prototype supports a planting sequence of hole formation, seed dispensing, and ground travel to another site or return to flight configuration.The drilling depth is configurable up to 30 mm, and the airframe weighs less than 500 g excluding the battery.Source: §I · p. 2

II. RELATED WORK

  • Unlike cited aerial-ground platforms, which demonstrate mobility or object manipulation, this work addresses soil drilling and seed embedding,,,,.The cited drone-seeding systems control release location but do not actively set burial depth [3],.Source: §II · pp. 2–3

III. PROBLEM FORMULATION AND DESIGN REQUIREMENTS

The task is to reach restoration sites by air, then plant at nearby locations without spending flight energy on each short hop. The design must reconcile flight, arm deployment, and drilling loads while meeting five hardware and control requirements.

III. PROBLEM FORMULATION AND DESIGN REQUIREMENTS

  • The robot must reach planting sites by air, drill unprepared ground, place one seed, and either travel to another nearby site or take off.The required tool reach is on the order of the 5–20 mm shallow-burial depths reported in, with appropriate depth depending on species and site.Source: §III · p. 3
  • The research question is whether one multirotor can provide aerial access, ground repositioning, and controlled-depth penetration without a separate actuator for each function.The design must reconcile high-speed, low-torque lift, brief high-reaction arm deployment, and sustained high-torque, low-speed drilling.Source: §III · p. 3
  • Five requirements cover flight-compatible mass, load-bearing rolling stance, controllable tool reach, unpowered pose holding, and isolation of flight-critical control.The stated geometry yields a maximum drilling depth of 30 mm; final seed depth also depends on how the seed settles in the borehole.Source: §III · p. 3

IV. SYSTEM OVERVIEW

The system transforms from a conventional multirotor into a four-point ground rover with an engaged drill. Its mission sequence combines flight, morphing, drilling, dispensing, and either ground repositioning or return to flight.

IV. SYSTEM OVERVIEW

  • The architecture uses the rear arms as ground contacts, a descending front stage to position the drill, and propulsion motors to drive it in plant mode.The sequence diagram identifies five states: flight, morphing to rover, drilling, dispensing, and morphing to flight.Source: §IV · p. 3
  • After landing, the rear arms fold onto their propeller guards and the front tray lowers to engage the drill gear train.The drill runs for a configurable 5–10 seconds; the tray is retracted and the seed gate cycles before the vehicle advances or returns to flight.Source: §IV · pp. 3–4
  • A dedicated ArduPilot flight controller handles flight, while a decoupled companion microcontroller runs morphing logic.Mechanism commands are kept separate from flight-critical control.Source: §IV · p. 4

V. ACTUATOR-MINIMAL MORPHING MECHANISM

The design minimizes added actuation by coordinating rear-arm deployment and using the propulsion motors as the drill drive. Mechanical stops and gearing support the deployed stance, while the front tray couples the drill only in plant mode.

V. ACTUATOR-MINIMAL MORPHING MECHANISM

  • The arrangement combines a coordinated rear-arm motor pair, a front-tray servo stage, and propulsion-motor reuse for drilling.The paper frames this as actuator reuse across flight, locomotion, and drilling.Source: A. Design Rationale: Actuator Reuse… · p. 4
  • The rear-arm motors share a control channel but drive mechanically independent gear trains; end-stops define terminal configurations and non-backdrivable gearing holds the arms without continuous current.The shared command does not guarantee equal motion under unequal loads.Source: A. Design Rationale: Actuator Reuse… · p. 4
  • A 10 mm front-tray descent meshes propulsion-motor shafts with the drill gear train, reusing flight motors to supply drilling torque.Separate flight and drill commands prevent simultaneous thrust and drilling; the tray can also tamp a seed that has not settled fully.Source: A. Design Rationale: Actuator Reuse… · p. 4

B. Primary Drive and Kinematic Coupling

The rear-arm drive couples two independent motors through one control channel, while mechanical end-stops define the deployed configuration. A descending front tray engages the propulsion motors with the drill train, and a separate gate dispenses seed.

B. Primary Drive and Kinematic Coupling

  • The electronically coupled dual-motor assembly drives one geared motor per rear arm from a single control channel, with each arm moving through its own rack travel.The arms reach their terminal deployment angles at mechanical end-stops; non-backdrivable gearboxes retain those configurations without holding current.Source: B. Primary Drive and Kinematic Coupling · p. 4
  • The shared drive signal does not guarantee equal arm motion, and synchronization under variable loading is not quantified.Differences in friction and load can produce unequal motion; the end-stops define the terminal configurations.Source: B. Primary Drive and Kinematic Coupling · p. 4

C. Front Tray and Drill Engagement

  • A 10 mm tray descent meshes the propulsion motor shafts with the drill gear train, reusing the flight motors and eliminating a dedicated drill drive.The tray stroke and drill-bit geometry set a maximum drilling depth of approximately 30 mm within the ground-clearance budget afforded by the propeller-guard wheels.Source: C. Front Tray and Drill Engagement · p. 4
  • After the drill retracts and the outlet aligns with the borehole, a geared motor opens the rack-and-pinion seed gate and then closes it.The seed-box floor slopes forward by approximately 15° to feed remaining seeds toward the outlet as the hopper empties.Source: C. Front Tray and Drill Engagement · p. 4
  • The mission sequence repeats drilling and dispensing at successive sites through a row-advance branch without re-entering flight.Morph transitions use a fixed drive interval, and mechanical end-stops define terminal configurations.Source: Figure 3 · p. 5

VI. MORPHING AND SEED-GATE CONTROL

A companion microcontroller issues separate commands for morphing, tray motion, and seed-gate operation, while the autopilot handles airborne stabilization. The rear-arm transition uses a fixed-duration command, and position-commanded actuators use smooth interpolation profiles.

VI. MORPHING AND SEED-GATE CONTROL

  • Airborne stabilization is delegated to a separate autopilot, while a companion microcontroller accepts independent rear-morph, front-tray, and seed-gate commands.Automatic triggering of those commands from autopilot landing and mission events remains to be integrated.Source: §VI · p. 5

A. Timing-Based Morph Sequence

  • The five-state sequence runs from FLIGHT through MORPHTOROVER, DRILLING, and DISPENSING to MORPHTOFLIGHT by issuing companion-controller commands in order.Integration with the autopilot’s landing and waypoint events is future work.Source: A. Timing-Based Morph Sequence · p. 5
  • The rear-arm drive is open loop and applies direction σ for a fixed 1.65 s interval before stopping.The interval is selected to carry both arms to their mechanical end-stops, but arm position is not measured during transition, so endpoint accuracy depends on reaching those stops within the allotted time.Source: A. Timing-Based Morph Sequence · p. 5
  • Position-commanded actuators follow a cubic ease-in/ease-out interpolation between end angles, starting and ending at zero velocity.The profile is realized in firmware as fine incremental interpolation to avoid the shock of a step command.Source: A. Timing-Based Morph Sequence · p. 5

B. State Persistence

The controller stores logical morph and tray states for firmware recovery, while the seed gate runs a timed opening-and-closing cycle. The drill-and-seed sequence aligns the outlet over the borehole before dispensing.

B. State Persistence

  • On each state transition, the microcontroller stores the binary morph or tray state in EEPROM and restores the last stored logical state at power-up.This does not verify physical pose; after an interrupted transition, the mechanism configuration must be checked before operation resumes.Source: B. State Persistence · p. 5

C. Seed-Gate Controller

  • The seed gate uses an 80 ms opening pulse, a 1 s dwell, and an 80 ms closing pulse, for a nominal 1.16 s cycle.Its sliding cover has two dispensing slots per cycle and supports seeds up to 10 mm in diameter; the printed insert can be swapped for other seed profiles without structural changes.Source: C. Seed-Gate Controller · p. 5, Figure 4 · p. 6

A. Drill-and-Seed Cycle

  • After the tray retracts, the vehicle advances until the seed-gate outlet sits over the borehole, then dispenses a seed.On loose or uneven substrate, a seed may fail to enter on the first attempt; the drill motion can be repeated so the drill tip tamps the seed into the hole before retracting.Source: A. Drill-and-Seed Cycle · pp. 5–6

B. Experimental Platform

The prototype combines aerial and rover operation with a drill-and-seed cycle, and the performance section reports flight capability, planting-cycle measurements, an illustrative mission-energy model, and a measured morphing response.

B. Experimental Platform

  • The airframe uses a 3D-printed PETG frame with carbon-fibre tube arms, and its propeller guards double as passive ground wheels.The prototype component list spans propulsion, autopilot, morphing, and seed handling.Source: B. Experimental Platform · p. 6

C. Performance

  • The motors produce 5.3 kgf static thrust on 4S, with a thrust-to-weight ratio above 6:1 at 0.8 kg all-up mass; hover endurance is approximately 10 min.The rover drives on 127 mm propeller-guard wheels.Source: C. Performance · p. 6
  • Drilling reaches approximately 30 mm, and tray descent, drilling, retraction, and seed dispensing take approximately 20 s per site.The sequence may repeat drilling to tamp a seed that does not settle into the borehole.Source: C. Performance · p. 6, Figure 5 · p. 7
  • With equal traversal and drilling times, the illustrative model gives 28.7 min runtime for 2 min of flight time, versus a 33.3 min ground-only limit.The assumed ground-operation runtime multiplier relative to continuous flight is 3.33.Source: C. Performance · p. 6
  • In one recorded drone–rover–drone cycle, the pitch proxy settled 5.07° from its initial reference in rover mode, while the reverse transition had a larger transient.The nominal 1.65 s drive interval is aligned to estimated onset in Fig. 6; this is not a command-to-completion measurement.Source: C. Performance · p. 6, Figure 6 · p. 7
  • During drone-to-rover reconfiguration, a pronounced gyroscope and acceleration transient occurred approximately 0.65–0.70 s after estimated onset, followed by leveling.The proposed rear-wheel-contact explanation and possible seed-transport effect are mechanical interpretations, not direct measurements.Source: C. Performance · p. 6

D. Morph Repeatability and Substrate Dependence

Five consecutive flight-to-plant-to-flight cycles completed deployment and retraction on a rigid bench using a 1.65 s drive interval.

D. Morph Repeatability and Substrate Dependence

  • Five consecutive cycles verified repeated endpoint reach on a rigid bench, but timing dispersion under load remains unquantified.The authors identify larger trials with arm-angle and motor-current measurements as needed to quantify that dispersion.Source: D. Morph Repeatability and Substrate… · p. 7

E. Seed Placement Trials

Outdoor trials showed seed entry into the borehole in 47 of 50 cycles, while the work remains a proof of concept with several field-use boundaries, including terrain, mission integration, and germination evaluation.

E. Seed Placement Trials

  • 47 of 50 outdoor drill-and-dispense cycles placed the seed in the borehole; germination was not evaluated.The observed success fraction was 0.94, with a 95% Wilson interval of [0.838, 0.979].Source: E. Seed Placement Trials · p. 7

VIII. DISCUSSION AND LIMITATIONS

  • Field deployment still requires broader terrain testing and integration of the mission stages beyond the bench morphing and outdoor placement trials.Source: §VIII · p. 7
  • Variable soil resistance during rear-arm deployment could desynchronize the rigid-bench-calibrated timing sequence from the mechanism’s true state.A current-sensing stall detector is proposed as a closed-loop extension.Source: §VIII · p. 7
  • Morphing and planting use separate companion-controller commands, and changing hopper mass may alter the center-of-gravity shift after the vehicle returns to flight.Triggering commands from autopilot landing and mission events is identified as a next integration step.Source: §VIII · p. 7
  • Wind and sloped or uneven terrain remain untested, and tray descent and drill engagement assume a roughly level landing attitude.Source: §VIII · pp. 7–8
  • Terrain mapping and arm-position or contact sensing could guide landing and deployment, but independent terrain conformation would require added actuation or mechanical compliance.Source: §VIII · p. 8
  • The study establishes seed embedding, but comparing germination and establishment with broadcast or seed-ball methods requires a separate field trial.Source: §VIII · p. 8
  • Repeated loaded-cycle tests are needed to quantify gear-train wear, backlash, and deployment-timing changes; candidate metal interfaces must be assessed against added mass and flight-dynamics effects.Source: §VIII · p. 8

IX. CONCLUSION

The conclusion presents a hardware-validated proof-of-concept robot for flight, ground repositioning, and controlled-depth seed embedding. Its ground mobility supports multiple planting sites in one sortie, while terrain-adaptive deployment remains a future direction.

IX. CONCLUSION

  • The physical prototype combines flight, ground repositioning, and controlled-depth seed embedding through a geometry-assisted, tri-functional morphing design.Source: §IX · p. 8
  • An integrated control system sequences flight, morphing, and seed dispensing using geometry-assisted timing and state persistence across power cycles.Source: §IX · p. 8
  • Ground locomotion on deployed rear wheels enables multi-site planting in one sortie, while LiDAR-guided deployment is proposed as a future extension.Source: §IX · p. 8
Loading 2609.37072v1…