Ropedia raises $22M to scale data collection for training robots

Ropedia, which is constructing information infrastructure for bodily AI, at present stated it has raised $22 million in pre-Sequence A funding. Mixed with the corporate’s seed spherical, it has raised $30 million.

The company plans to make use of the funding to scale HOMIE, its wearable, head-mounted machine. HOMIE captures first-person human motion, object interplay, and spatial context after which feeds that information into Ropedia’s processing and annotation fashions.

“For this spherical of fundraising, we’re actively increasing our enterprise and technical networks within the sense that we wish to construct up our groups in {hardware}, software program, information infrastructure, and mannequin growth,” Zhaoxi Chen, co-founder and CEO of Ropedia, instructed The Robotic Report. “The second utilization is de facto going to be increasing our enterprise to North America, particularly in america, as a result of nearly all of our present purchasers and prospects are primarily based there.” 

Ropedia additionally plans to construct out its information platform, including annotation tooling, high quality analytics and compliance infrastructure, and to develop its AI analysis group’s work on information, basis fashions, and world fashions. Based in 2025, the corporate has places of work in Singapore and Mountain View, Calif.

The robotics trade wants information infrastructure

The information pipeline that begins with HOMIE runs finish to finish, from seize to model-aligned fine-tuning. This creates information that Ropedia generates and buildings itself fairly than the client-owned information that the majority data-labeling suppliers annotate. That full-stack method additionally units it aside from teleoperation-based seize, which requires bodily robotic {hardware} and is usually restricted to particular robotic embodiments.

Chen asserted that Ropedia is constructing the info infrastructure the robotics trade wants if it desires to scale.

“Infrastructure means you’re going to have the ability to course of large scale of knowledge in a short while,” he stated. “Delivering 1,000 hours of knowledge is a completely totally different factor in contrast with delivering 1 million hours of knowledge, as a result of it’s worthwhile to ensure that all the things is trade stage of requirements.”

What’s HOMIE?

HOMIE is a data-collection machine geared up with 360º digicam view. Ropedia goals to construct human-level intelligence in robotics, and to do that, Chen claimed that one of the best place to start out is with copying human conduct.

“You’re going to imitate human-level intelligence by going again to the best way people study new ability units,” he stated. “We study within the bodily world via our imaginative and prescient, via our interplay from a first-person perspective. We study from a first-person view. We study all bodily expertise or high-level reasoning semantic ability units from this attitude. So, that’s why we imagine within the energy of selfish information.”

HOMIE consists of different sensing modalities along with imaginative and prescient. “We even have an audio sensor the place you may reconstruct the supply and course of an audio sign within the setting,” Chen stated. “We even have a movement unit on the headset the place you may monitor how you progress.”

A part of what helps Ropedia stand aside, Chen stated, is that the corporate developed a lot of HOMIE in-house. This tradition {hardware} leads to higher information.

“We’re not a very full-stack firm, however the majority of the {hardware} we’re growing purely on our personal,” he famous. “We don’t develop CMOS sensors, resistors, and people sorts of off-stream elements. However the total buildings, the PCBA, and the {hardware} synchronization of various sensors, we’re growing on our personal.”

Ropedia stated its method cuts data-collection prices by as much as 50 instances in contrast with conventional strategies. | Supply: Ropedia

What duties is Ropedia focusing on?

In the long run, Ropedia is focusing on a variety of duties for its datasets. “Ideally, we wish to seize all the duties, however we have to do issues step-by-step. We’re beginning with seize in an off-the-cuff setting within the sense that we seize the info within the house,” Chen stated.

To this point, the corporate has been specializing in duties like housekeeping, cleansing, and even inside design.

“Ideally, we wish to seize all the duties, however we have to do issues step-by-step. We’re beginning with seize in an off-the-cuff setting within the sense that we seize the info within the house,” Chen stated. “We even have a pickleball membership the place we require all of the members to put on our headsets and do the recording throughout their follow.”

Proper now, Ropedia collects round 1,000 hours of knowledge per week, with 100,000 whole hours gathered. The corporate stated it’s already offering worth to its early prospects.

“What we have now been doing is offering two kinds of providers. One is OTS or off-the-shelf datasets. We’re instantly licensing our present information within the database, and they’ll use our information on demand to meet their analysis and foundational consumption,” Chen stated. “The second sort of service for our prospects is we purchase the high-quality information.”

Save the date for RoboBusiness 2026

What are Ropedia’s subsequent steps?

Ropedia intends to proceed increasing information assortment in Southeast Asia and North America, rising its U.S. group, and rising manufacturing of HOMIE. The corporate additionally plans to collaborate with upstream and downstream companions and is taken with working with trade, academia, and even robotics hobbyists.

Trying forward, Ropedia plans to create extra wearable information seize {hardware}, stated Chen.

“There’s a loopy plan in the way forward for Ropedia the place a human being goes to function a socket, within the sense that we wish to put as many gadgets on high,” he stated. This might embody the headset in addition to gadgets on the palms or legs to seize extra information.

“Now we have been growing the following step of the product line. So, one goes to be a hand machine the place you’re precisely monitoring the hand pulses and tactile interplay,” Chen stated. “On the identical time, it’s going to synchronize with our HOMIE gadgets, so you may have synchronized information streams.”

Ropedia additionally has plans for different much less standard gadgets, together with sneakers that might sense how ft work together with the bottom to assist prepare humanoid robots.

The publish Ropedia raises $22M to scale information assortment for coaching robots appeared first on The Robotic Report.