Unkyoung Won
Team lead
Move your cursor and the arm follows. Click a box to move that box, or a yellow docking port to send the arm there.
Every grip becomes a path.
D-end’s symmetric docking manipulator separates the base it travels on from the base it works from. Put identical docking ports on ceilings, walls, floors, workbenches, AMRs and quadruped robots, and the arm moves itself from port to port. Built by four robotics engineering students at Hanyang University ERICA.
The arm was never the problem. The feet were.
A manipulator does its work through the end effector at its tip. To reach the work point it is either installed in a fixed position or carried there on an AMR or a biped or quadruped platform. But because the arm and the mobile platform share a single base, every constraint of that platform transfers straight through to the arm’s work.
-
Access
If the AMR cannot clear a door sill or a cable tray, the arm cannot enter that workspace either.
-
Reach
For a fixed arm, where it stands decides what it can work on. On a wide bench, anything outside the working envelope is simply out of the end effector’s reach.
-
Precision
Reaction forces from the work travel back into the mobile platform. On biped and quadruped platforms the body’s posture control and the arm’s control have to be handled at the same time, which makes control for precision work far more complex.
Resolving any of this the conventional way meant bringing in a different robot for every environment, or heavily rebuilding the existing workspace around the robot.
Both ends are identical. That is why it goes anywhere.
We propose a docking manipulator that separates the base used for travel from the base used for work. Rather than improving the vehicle, we made the arm detachable — an approach taken from Canadarm2, which moves across the International Space Station by docking end over end. The arm carries the same docking interface at both ends, and ports of the same standard sit on mobile robots and on fixed work references alike.
Ride a platform to the work area
The arm travels docked to an AMR or a walking platform until it reaches the entrance of the work area.
Latch the free end onto a site port
Only after that latch is confirmed does the platform-side dock release.
The vehicle leaves, the arm works
The released vehicle heads off to its next job while the arm stays fixed to the port.
Out of reach? Take the next port
When the working envelope runs short, the arm hands itself over to a nearby docking port, inchworm style, and repositions on its own.
Access by changing vehicles
The arm swaps vehicles, so it enters a workspace regardless of whether any one vehicle can get in.
A working envelope that relocates
When something is out of reach, the arm moves to a port near the work point and redraws its envelope. Add a port instead of building a longer arm.
Precision on solid structure
This problem only exists on platforms that are never fully fixed. The arm walks onto rigid structure itself and gains stability from it.
Deployment without renovation
No large-scale rebuild and no starting from scratch. Install ports where they are needed and bring whichever vehicle suits the environment.
In that sense our docking port is not simply a fixture. It works as a standard adapter between a wide range of vehicles and the arm.
One arm. Standard ports. Interchangeable hands.
The system consists of a symmetric manipulator body, standard docking ports and modular end effectors. Ports come in the five forms below, unified to one standard so they can be placed to suit the site. The arm moves between them on its own and works from a fixed base at each work point.
- Ceiling
- Wall and floor mount
- Workbench mount
- AMR mount
- Quadruped mount
Docking latch mechanism
A hook-type lock that will not release even if power is cut suddenly, guarding against a drop. The mouth is wide and narrows inward so actuator error during latching is absorbed, and the play inherent to hook latches is taken up by a dedicated locking actuator and the mechanical design.
Port detection and alignment
ArUco markers and a camera identify the port, and the position and orientation error that builds up while approaching it is corrected.
Base frame redefinition
Once docking completes, the kinematic reference switches using the port ID and its pre-registered coordinates, with an IMU verifying the orientation of the mounting surface.
Redundant latching
The order is enforced: the existing latch releases only after the opposite latch is confirmed. Latches are doubled so that no single failure ends in a drop.
Leave the benches alone. Just add ports.
Conventional automation often required changing the work environment itself to suit the robot. That meant touching the structure of the site, and paying for it. So smaller facilities — the ones mobile robots cannot enter, or the ones without the capital for a fleet — have had high demand for automation and little practical way to adopt it.
Our entry market is laboratory automation. Work points are fixed and rarely move, which suits port placement, and it is exactly the kind of space existing mobile robots struggle to enter.
Floors that cannot be touched
Lab floor structure is hard to change. Ports attach to existing benches and walls, so the space never has to be rearranged.
Fixed workstations
Existing lab automation equipment is installed at one specific workstation, which means the bench layout has to change along with it.
High positional accuracy
Reagent dispensing, plate transfer and instrument loading all demand tight positioning. Working from a fixed port meets that requirement.
Frequent tool changes
Pipettes, grippers and cap openers are swapped from task to task. Unifying base latching and tool changing into a single standard pays off directly here.
We start with university labs and small corporate research labs that cannot justify dedicated automation equipment.
A foothold when moving. A base when working.
| Approach | Strength | Limit |
|---|---|---|
| Fixed collaborative robot | Precision | Working range tied to the installation point |
| Mobile manipulator | Can move | Floor obstacles and aisle width |
| Biped and quadruped | Excellent access | Payload and precision control |
| D-end | Separates the travel base from the work base, securing mobility and working stability at once | |
If the table looks cut off, scroll it sideways.
A vehicle chosen per environment
An AMR in the open, and a walking platform where the AMR cannot enter. Once inside the work area the arm moves between ports on its own and extends its working range.
Payload and precision on demand
When more lift or more accuracy is needed, the arm moves onto a fixed base, or onto an AMR, which is steadier than a walking platform.
Expansion that stays cheap
More work points or a changed layout does not call for another robot. Add or relocate ports and one body does the work of several.
One unified interface
Base latching and tool changing share a single standard, so moving and swapping tools run through the same mechanism and automate together.
To hold these as our own technical advantage, the docking mechanism and the correction control method are planned for patent filing.
Modular hardware. Recurring revenue.
The modular split between body, docking ports and end effectors carries straight into the revenue model.
Hardware package
Body, ports and end effectors sold as a package to secure initial revenue.
Module add-ons
Ports and end effectors sold individually as modules, creating repeat revenue.
Monitoring subscription
Docking position error, joint load and retry counts monitored live without extra instrumentation, with anomaly detection and predictive maintenance on top.
B2B — integration contract
Body, ports and end effector set delivered as one package. Site survey, port layout design and initial setup are included in the build contract.
B2C — maker kit
A self-install kit with no build contract, with ports and end effectors sold separately so users can widen their working range step by step.
The B2C channel exists for application cases and data more than for revenue. The dimensional specification and basic control of the low-cost docking interface are published; the latch implementation and the docking correction algorithm are not. Cases built by users become both a reference for B2B sales and a route to discovering new applications.
From proof to rollout. Four stages.
Base handover demo
Build the standard port and docking interface and install them to the same standard on an AMR deck, a wall and a workbench. Demonstrate one arm moving through the three points repeating the same task, and present docking repeatability, time per dock and payload as numbers.
Self-relocation
Complete the symmetric ends so the arm crosses between ports on its own, and demonstrate the full scenario of moving between ports mounted on different surfaces such as ceiling and wall.
Field validation
Run a pilot at one site, then check and refine port placement rules, real working accuracy and the problems that appear in operation.
Standardization and rollout
Fix the port and end effector specifications and expand to other sites in the same industry.
The people building it.
Four of us studying robotics at the Department of Robot Engineering, Hanyang University ERICA.
Gwangmin Nam
Software
Sihyun Lee
Software
Jiwon Jang
Hardware