Robot Delivery in Seoul: Where Things Stand

A Robot Actually Delivered My Food

Last month in Gangnam, I ordered lunch and a robot showed up. It was a Baemin Dilly robot -- a boxy unit about 60cm tall that crept along the sidewalk. You enter a verification code in the app, the lid opens, and you grab your food.

It was cool, but delivery took about 47 minutes. A human rider would've covered the same distance in about 18 minutes. (At this speed, I should probably stick to ordering cold dishes.)

Current Status in Seoul

As of June 2026, delivery robots operate in about 8 districts in Seoul: Gangnam, Seocho, Songpa, Mapo, Seongdong, and others. Most deployments are concentrated around apartment complexes and office districts.

Baemin, Coupang Eats, and Neubility are the three operators, with a total fleet of about 370 robots. Robot deliveries account for about 0.3% of Seoul's delivery market. It's still very much experimental.

From what I can see, this number needs to reach at least 5,000 units to become meaningful. At the current growth rate, that's probably 3-4 years out.

The Technical Side

The core technology behind delivery robots is essentially the same as autonomous driving. LiDAR, cameras, GPS, and IMU sensors work together to perceive the environment and plan routes.

But there are challenges that car-based autonomous driving doesn't face. These robots have to navigate sidewalks -- pedestrians, bicycles, electric scooters, merchandise stacked outside shops, illegally parked cars. Sidewalks are far more unpredictable than roads.

I've actually seen a photo of a delivery robot stuck on a curb. A height difference of about 3cm that its wheels couldn't climb, requiring a remote operator to intervene. Seoul's irregular sidewalk infrastructure is a major obstacle.

Cost Structure

Human rider delivery costs about 3,500-5,000 won per order. Robot delivery currently costs about 4,200 won per order, including robot depreciation and maintenance.

It's still more expensive than a rider. But robots can work 24 hours (minus charging time) and their labor cost doesn't increase over time. Once delivery volume exceeds a certain threshold, robots become cheaper. The break-even point is estimated at about 23 deliveries per day.

What's missing from this calculation is remote monitoring labor costs. Currently, each remote monitor oversees 3-4 robots. Until full autonomy is achieved, human labor costs remain.

What Happens to Riders?

This is the most sensitive issue. There are about 500,000 delivery riders in Korea, and the question is how many of those jobs robots will absorb.

Honestly, the short-term impact is negligible at 0.3% market share. But looking 5-10 years out, things could change. If robots handle 30-40% of simple deliveries, rider demand will inevitably shrink.

That said, not all deliveries can be replaced by robots. Stair deliveries, unpaved areas, urgent orders. Some things only humans can handle. Riders won't completely disappear, but the number of jobs will decrease.

My Take

Technically fascinating, but social acceptance needs time. Robots roaming sidewalks still feels a bit odd. Pedestrian right-of-way issues, accident liability, noise concerns -- all still need to be sorted out.

But five years ago, when shared scooters first appeared, people also said "will this really work?" and now they're part of daily life. Delivery robots might follow a similar trajectory. Just probably on a longer timeline.