The Unavoidable Convergence of Imaging and Robotics Technology

"How exactly will robots integrate into our daily lives?"

This question has likely been one of the most recurrent inquiries within the embodied AI sector in recent years. Some advocate for robots to first master household chores and elderly care, eventually evolving into versatile household aides. Others argue that instead of waiting for an all-encompassing humanoid robot, it's wiser to have robots excel in a specific task initially.

For instance, assisting in capturing photos and videos.

In July this year, Insta360 unveiled its long-term vision for a "photography robot" dubbed Cameraman, aiming to develop a Camera Agent capable of autonomously perceiving its surroundings, locating optimal shooting positions, framing shots, and tracking subjects, without being confined to any particular product design.

Image Source: Insta360

Simply put, Insta360 envisions a camera that operates independently, without constant human intervention, and can accomplish shooting tasks autonomously.

Around the same time, Miaodong Technology, founded by a former DJI technical expert, launched Beni—a compact, two-wheeled robot resembling a pet. It can follow users, navigate obstacles, jump, and capture 4K videos.

Image Source: Miaodong Technology

Compared to Insta360's Cameraman, which is still in the exploratory phase, Beni has already completed crowdfunding, with early bird prices starting at $549 (approximately RMB 3,674) and is set to begin shipping in November. Domestic sales plans are aligned with large-scale overseas distribution, meaning it's just:

One step away from reaching consumers.

And it's not just these two companies eyeing "photography robots." DJI and Hover have long explored autonomous tracking through drones, Qiyuan's T1 and VITA Dynamics' robot dogs have started handling mobile photography tasks, and even Honor has integrated mechanical gimbals and AI Agents into its Robot Phone. The difference is that some create robots specifically for photography, while others aim to teach robots how to photograph.

As the saying goes, all roads lead to Rome. But in the journey of embodied AI reaching ordinary users, will "photography" be a narrow path or a broad avenue?

Letting the Camera Move Autonomously

If we consider photography robots as devices capable of autonomous movement and tracking users for shooting, then they have already entered the consumer market, albeit rarely referred to as "robots" in the past.

The most typical example is drones. Take the DJI Neo 2, for instance. This lightweight drone, weighing only 151 grams, supports palm takeoff and landing, gesture control, intelligent tracking, and omnidirectional obstacle avoidance. Users don't need to constantly hold the remote control for it to follow them while running, cycling, or performing cinematic maneuvers like orbits and farewell shots.

Zero Zero Robotics' Hover Air (HOVERAir) even launched the auto-shooting flying camera X1 as early as 2022, followed by products like the X1 Pro and X1 ProMax. Compared to traditional drones, they emphasize their identity as "flying cameras" rather than "drones," with palm takeoff, auto-tracking, preset cinematic moves, and minimizing flight operations during shooting.

X1 ProMax, Image Source: HOVERAir

In a sense, these products already possess the fundamental characteristics of photography robots: they can perceive targets, control their movement, and, to some extent, autonomously complete shooting.

Of course, flying has its limitations. Battery life, noise, usage venues, and safety restrictions all determine that drones cannot cover all shooting scenarios. Especially in indoor environments or crowded streets, ground-moving robots may have an advantage.

This is also a key reason why Miaodong's Beni opted for a two-wheeled, legged design.

Unlike drones shooting from above, Beni can follow users on the ground, using its two wheels and joint structure to turn, jump, and adjust its posture. It can capture 4K videos, such as dynamic shots close to the ground of pets running, children playing, or skateboarding, and can also express a pet-like sense of interaction through different movements.

TechRadar was impressed by Beni's interactive design, believing that its expressions, sounds, and rolling actions make it resemble a lively robot pet.

More manufacturers are starting to explore this approach. For example, Qiyuan Robot's T1 not only features a transformable design that switches between two wheels and four legs but can also connect to action cameras like those from Insta360, supporting voice shooting, auto-tracking, and orbital cinematic moves. Vbot's Super Robot Dog, besides autonomous tracking, obstacle avoidance, and carrying objects, can also mount action cameras to serve as a mobile photographer during outdoor activities.

Qiyuan T1 connected to Insta360 Luna Ultra, Image Source: Qiyuan Robot

Compared to Beni, these two products are not designed exclusively around photography. Photography is just one of the important tasks they can perform, with the robots themselves carrying more functions. This difference is quite intriguing.

Even Honor's Robot Phone, launched this year, directly integrates a mechanical gimbal into the phone. Through AI subject recognition, auto-tracking, and pre-composition, the lens can actively adjust its direction as the subject moves, even completing some professional cinematic moves.

Similarly, the Obsbot Tail 2 achieves subject tracking through its AI-active gimbal design and supports a set of highly reference-worthy gesture interaction designs (for details, see Leitech's review "Obsbot Tail 2 Live Streaming Camera Review: An AI Gimbal Camera and Also an Imaging Robot"):

The active gimbal of the Obsbot Tail 2 doesn't mean we can casually place the camera in any "obscure" position: For convenience in single-user operation, the Obsbot Tail 2 has gesture control capabilities: forming an "L" shape with the thumb and index finger can control zooming, while the "OK gesture" can start or stop shooting. However, if you don't place the Obsbot Tail 2 facing forward, the camera may not accurately recognize the "OK" gesture.

Image Source: Leitech

There have been cameras with tracking capabilities before, but no matter how intelligent the algorithm is, it can only adjust image parameters, not the position and angle of the lens. Broadly speaking, both Honor's Robot Phone and Obsbot's Tail 2 can be considered a type of "photography robot":

At least they prevent the camera from passively waiting for user operation and instead give it a certain degree of physical autonomy.

As for Insta360, although Cameraman currently lacks a definite independent product form, it has already accumulated a wealth of relevant technology through panoramic cameras, Flow phone gimbals, Luna gimbal cameras, and the Link panoramic drone. Insta360's vision is not to create a robot in a specific form but to enable the camera to understand user intent and autonomously complete shooting, editing, and even sharing.

It must be admitted that today's "photography robots" are hard to define by a single product form. They can fly in the sky, run on the ground, resemble a robot dog, or even just be a set of mechanical gimbals behind a phone.

In a public interview, Zero Zero Robotics founder Wang Mengqiu said, "The camera flying is just a means, not the end," and Insta360's Liu Jingkang also mentioned when discussing Cameraman that "what most people really want is not to operate the machine but to get good photos and videos."

Photography robots are not a sudden new concept but the result of the continuous autonomy of imaging equipment and the gradual practical application of consumer-grade robots, the intersection of two technological paths. Ultimately, from a narrow demand perspective, the "robot" in "photography robot" can only be a means, not an end.

But at the same time, photography robots can serve as a gateway for robots to enter human daily life.

Learning to Track Before Becoming All-Powerful?

In many people's views, the ultimate form of embodied AI seems to be humanoid robots.

Miaodong Technology CEO Yang Shuo also doesn't believe that all robots should develop toward a humanoid form. Yang once participated in and led the development of DJI's flight control system, later graduated with a Ph.D. in robotics from Carnegie Mellon University, and served as an architect for Tesla's Optimus humanoid robot project.

He gave an example in a previous interview: If a user wants a robot to hold an Osmo and follow them around to shoot, why must we first build a humanoid robot capable of walking on two legs and holding a camera with its hands? A small vehicle or even a simpler mobile device could also accomplish the same task.

In fact, this group of Shenzhen manufacturers, including Insta360, Hover Air, DJI, and Miaodong, have a more pragmatic focus, emphasizing the satisfaction of "user needs" and valuing commercial implementation.

So it's not hard to understand why Miaodong chose Beni's current product form. It doesn't need to imitate the human body or master complex dexterous operations. As long as it can stably follow users and adjust shooting positions, it has a chance to complete photography tasks.

Image Source: Miaodong Technology

Another point is that photography requires robots to perceive their surroundings, recognize shooting subjects, plan their own movement routes, and continuously adjust their actions based on the actual situation, but it doesn't necessarily require robots to directly operate and change external objects. This means that compared to complex household chores like tidying up, doing laundry, and cooking, photography can to some extent avoid technical challenges such as dexterous hands and object manipulation while still allowing robots to truly enter complex and ever-changing real-world environments.

But this doesn't mean Yang Shuo is not optimistic about humanoid robots. On the contrary, Yang mentioned in his personal column that Miaodong Technology decided from the beginning to develop two product lines: "One is a more long-term humanoid robot, and the other is a smaller, safer, and easier-to-enter-real-life small robot."

In fact, the advantage of humanoid robots lies in versatility. The environments, tools, and even various infrastructures in human life are mostly designed around the human body. If robots need to operate these tools and complete various tasks, a humanoid structure naturally has its value.

But if only one specific task needs to be completed, the situation is different. For example, a robot following someone to shoot doesn't need two legs, two hands, or to first learn to tidy up, do laundry, and cook.

The key lies in being "smaller, safer, and easier to enter real life first." From the perspectives of technical maturity, price, practicality, and product interaction, a product form like Beni clearly has the opportunity to enter mass life earlier than humanoid robots.

Over the past decade-plus, phone, camera, and drone manufacturers have been lowering the barrier to photography. From autofocus and computational photography to intelligent composition, auto-tracking, and now AI editing, more and more tasks that originally required users to do themselves have been handed over to devices.

Following this path, the next natural step is to let the device decide how to shoot and even when to shoot on its own.

In fact, in Liu Jingkang's vision, the core of Cameraman is a Camera Agent capable of understanding user intent, autonomously selecting lenses, shooting positions, and compositions, and ultimately delivering satisfactory images. In other words, the development of photography robots is not just about teaching devices to move but also about gradually helping them understand photography itself.

Another route starts from robotics.

Today, the problem with robots entering life, besides cost and other factors, is more about "what can they do." Martial arts and dance performances, making ice cream, coffee, pancakes, stocking shelves, and retail—these robots are mostly geared toward businesses. Relatively speaking, "photography" is more personal, a point also validated in consumer-grade drones, handheld gimbal cameras, and action cameras.

Image Source: Leitech

So for robots, photography can be a core function or just one of many abilities. Robots can be designed specifically for shooting or simultaneously handle tracking, companionship, carrying objects, and other life services, thereby attracting consumers to purchase.

These two routes are not mutually exclusive. They start differently but are intersecting in the specific scenario of photography. This also reminds us that the implementation of embodied AI doesn't necessarily need to wait for an all-powerful general-purpose robot.

However, having auto-tracking and obstacle avoidance functions is not enough. For photography robots to truly approximate a photographer, they also need to understand shooting timing, scene changes, and even user preferences for images.

So from Qiyuan's T1 to Vbot's Super Robot Dog, both choose to achieve tracking by mounting external cameras, letting manufacturers like Insta360 and DJI handle the actual "shooting" part, while the products also offer tracking, companionship, carrying objects, and other life services to attract consumers.

These exploratory endeavors have brought us one step closer to the consumerization of embodied artificial intelligence (AI). The ultimate form it will take—whether a drone, a wheeled-legged robot, or an advanced iteration of existing cameras and smartphones—is not the most critical consideration. What truly holds significance is whether robots can accomplish tasks that previously necessitated human involvement or were even impossible to perform independently, all without imposing an excessive usage burden.

This observation also prompts a more pragmatic inquiry. The demand for photography is undeniable, yet today's smartphones, panoramic cameras, drones, and intelligent gimbals are already capable of meeting a wide array of shooting scenarios. Do we genuinely require a new robotic solution?

Before the advent of practical photography robot products, I do not yet have a definitive answer.

Source: Leitech

All images in this article are sourced from: 123RF Royalty-Free Image Library. Source: Leitech