Computer Vision
7 min read
Computer Vision in Robotics: How is Computer Vision Used in Robotics in 2023?
Written by
Chinar Movsisyan

As we venture into the vibrant world of computer vision in robotics, we unlock a realm of possibilities while confronting intricate challenges. By bestowing robots with the ability to 'see,' we empower them to navigate and interact with their surroundings with remarkable autonomy. From facilitating intricate manufacturing processes to advancing autonomous vehicles and drones, the applications are as diverse as they are transformative. Yet, deploying computer vision in real-world robotics is not without hurdles. Achieving reliable object recognition in varied lighting conditions, angles, or cluttered environments remains a demanding task, and questions about privacy and model bias add layers of complexity. Nonetheless, as we tackle these challenges head-on, the exciting opportunity unfolds: a future where robots not only 'see' but also comprehend and deftly manipulate their environment, catapulting us into an era of unprecedented innovation.

Basics of Computer Vision in Robotics

In essence, computer vision refers to the ability of computers or machines to perceive, process, and understand visual data from the surrounding environment. Computer vision in robotics has become an indispensable tool that empowers robots to interact with their environment effectively. It endows machines with the visual sense, somewhat akin to the human capability of seeing and understanding visual data, thereby making them 'aware' of their surroundings.

This heightened awareness, made possible by computer vision, has found a myriad of applications in robotics. For instance, computer vision is extensively used in industrial robotics for tasks like quality inspection, where robots are used to visually inspect and analyze products for defects. In the realm of autonomous vehicles, computer vision helps self-driving cars recognize traffic signs, pedestrians, other vehicles, and more to navigate safely. It is also used in agricultural robots for crop monitoring and in healthcare robotics for precision surgeries.

Computer vision has significantly expanded the utility and effectiveness of robotics. It has transformed robots from simple, programmable machines to intelligent devices capable of making decisions based on their visual perception. This heightened intelligence allows robots to carry out tasks with a higher degree of accuracy, autonomy, and efficiency, extending their usability beyond controlled environments to a wide range of complex, real-world scenarios. Hence, computer vision in robotics has truly revolutionized the industry, propelling it towards an era of autonomous, intelligent machines.

What is the Difference Between Computer Vision and Robot Vision?

Two terms that are often conflated with one another are computer vision and robot vision. ‍Computer vision, a subset of artificial intelligence, aims to replicate and surpass human visual perception capabilities in machines. The objective is simple: enable machines to 'see,' interpret, understand, and then respond to visual data. Computer vision software uses algorithms to process, analyze, and understand images or video data. It can be used to identify objects, track movements, or even decipher handwriting. It might sound like the stuff of sci-fi, but in truth, we encounter applications of computer vision daily. They range from facial recognition in our smartphones, automated license plate readers, or AI-assisted medical imaging diagnostics.

Robot vision, on the other hand, can be considered a specialized application of computer vision. It is employed to equip robots with the ability to perceive, understand, and navigate the world around them. In essence, robot vision is the system that allows robots to 'see.' It uses cameras (the robot's 'eyes') and computer vision algorithms to process and interpret the visual data. The robot then uses this understanding to interact with its environment, whether it's picking up an object, avoiding obstacles, or performing complex assembly tasks in a factory.

But then, what truly differentiates the two? Simply put, the distinction lies in the use case. Both computer vision and robot vision involve interpreting visual data, but robot vision takes it a step further. It uses the interpreted data to guide physical actions in a real-world environment. Robot vision extends computer vision's capabilities by adding the element of physical interaction with the environment, a necessity for tasks requiring dexterity and spatial understanding. This is why robot vision is the lynchpin in areas such as autonomous vehicles and industrial automation.

There's an inherent symbiosis between computer vision and robot vision. The former serves as the foundation upon which the latter is built. Without the ability to interpret visual data (computer vision), a robot wouldn't be able to physically interact with its environment (robot vision).

Computer Vision Techniques for Object Recognition and Detection in Robotics

Object recognition and object detection are two of the most common tasks in computer vision, and they are crucial components in the automation landscape, particularly in the robotics realm. While they might appear synonymous, each carries distinct implications and applications.

Object Recognition is the process of identifying a specific object in an image or video. It's comparable to how humans recognize familiar items — for instance, spotting a favorite coffee mug amidst an array of dishes. In computer vision, object recognition often involves categorizing the object into predefined classes. For instance, a computer vision system might be trained to recognize and classify objects such as cars, pedestrians, or street signs.

Object Detection, on the other hand, is not just about identifying what an object is, but also determining its location in an image or video. It usually involves producing a bounding box around the object of interest, indicating where in the image the object is found. If object recognition is akin to recognizing your coffee mug, object detection is about pinpointing exactly where that mug is on your cluttered kitchen counter.

In the context of computer vision in robotics, both object recognition and object detection are vital. These abilities allow robots to understand and interact with their surroundings. For example, a robot in a warehouse might use object recognition to identify different items it needs to pick up and object detection to locate where those items are on a shelf. Autonomous vehicles leverage these capabilities to identify and locate other cars, pedestrians, and road signs, facilitating safe navigation.

Several cutting-edge computer vision techniques power object recognition and detection. Convolutional Neural Networks (CNNs) are a popular choice, known for their efficacy in image-related tasks. CNNs can learn and extract hierarchical patterns from raw image data, enabling them to identify and classify objects effectively.

YOLO (You Only Look Once) is another widely used technique, particularly in real-time object detection. Unlike other methods that might process an image multiple times to identify and locate objects, YOLO accomplishes both in one fell swoop, making it a fast and efficient choice.

Finally, Region-based Convolutional Neural Networks (R-CNNs) and their derivatives (Fast R-CNN, Faster R-CNN) are prevalent in both object recognition and detection. They work by proposing regions where an object might be located and then using CNNs to classify those objects.

Applications of Computer Vision in Robotics


Robotics has long been a point of interest for the manufacturing industry. Manufacturing often involves repeatable processes which robotic systems can conduct efficiently and effectively. These tasks involve material handling and pick and place tasks (sorting, palletizing, stacking, etc.), which are handled using robotic arms. Incorporating robotics into a company’s manufacturing process increases productivity and saves on labor costs. It can be expected that this trend will only continue as robotics and AI technologies continue to advance.

Distribution & Supply Chain

As e-commerce continues to take up a greater and greater share of consumer spending, distribution and supply chain challenges need to be optimized in order to meet the demands of the market. Robots are increasingly being used in warehouses to boost overall productivity and create more efficient workflows. Several companies today build what are known as autonomous mobile robots. These robots are able to learn their surroundings and plot routes to move around the warehouse. They can be used for moving pallets and products from their picking locations to the shipping or staging area. According to Honeywell, one of the world’s largest developers of software for process manufacturing, picking times can be reduced by “nearly 50%” by incorporating mobile robots into the workflow.

Increasingly, there are also several companies making advancements towards autonomous systems for trucks. Self-driving trucks will eventually significantly disrupt the supply chain process, as they will be able to work without the constraints that human-operated trucks have. Companies such as Tesla, Waymo and Torc are all actively developing autonomous trucks.

Robotics for Recycling

Robots are also being utilized in the recycling industry to conduct waste sorting. The United States alone produces approximately 300 million tons of trash a year. The portion of this that is recycled needs to be sorted, as not all pieces that are sent for recycling are actually recyclable. Traditionally, this job has been done by humans, who often have to work on a conveyor belt sorting through trash for items that can be recycled. As you can imagine, this is a challenging job, and often recycling centers can’t recruit enough people necessary to do their job. By utilizing computer vision in robotics, companies can fill the gaps to do this job. Computer vision models can be trained to detect and sort through the recyclable objects on a conveyor belt. The system will make a determination as to whether or not something is plastic, glass, paper—or something that is not recyclable—and sort them accordingly.

Real-World Challenges and Biases in Computer Vision for Robotics

In order for robots to be useful in dynamic environments, they must have some understanding of the tasks they’re trying to accomplish. In the distribution and supply chain examples we discussed earlier, the mobile robots that operate on the warehouse floor use computer vision to detect and track objects that may be blocking their way, and adjust accordingly. As the robots are moving boxes onto a pallet, they must be able to detect whether the pallet is within reach, if there is room on it, and where to stack the object onto the pallet. All of this is done using object detection and tracking algorithms.

A significant but often overlooked issue within computer vision in robotics is the potential for computer vision bias in the trained models. Machine learning algorithms, including those used in computer vision, are as good as the data they are trained on. If the training data is not diverse enough,his can lead to an algorithm that performs well under specific conditions but fails under others. For instance, in a warehouse context, if a computer vision model was only trained using data with boxes of a particular size or color, it might struggle to accurately identify boxes that deviate from these characteristics. These biases can limit the flexibility and robustness of the system in dynamic environments, and even lead to unintended negative consequences if certain objects or conditions are consistently overlooked or misinterpreted. It's therefore crucial to ensure diverse, comprehensive, and well-balanced data sets for training to build an unbiased and robust computer vision model.

In the example of robots that help sort through heaps of materials sent for recycling, the computer vision model must be able to identify cans as recyclable aluminum materials. While it may be easy to build a model that can identify soda cans on a shelf in a grocery store, the cans that end up on a conveyor belt in a recycling center can come in a variety of shapes. Some of them will be crushed to save space, others may be cut open or be significantly dented. Being able to correctly identify the object, regardless of its current shape, is an important aspect for the model to be effective in production. There is also the issue of occlusion, meaning that parts of the object that the model is trying to detect is not visible, usually because it is covered by something else. The position of the object can also differ from scenario to scenario, as can the position of the camera the robot is operating with. This can lead to some complexities for the computer vision algorithm to correctly detect the object.

Future Possibilities of Computer Vision in Robotics

As we peer into the future of robotics, one thing is clear: computer vision will play a monumental role in shaping this field. The advances we've seen so far — object recognition, object detection, image segmentation — are just the beginning. The horizon is filled with opportunities, and while we can't predict every breakthrough, several promising trends are emerging.

Advanced Human-Robot Interaction: With advancements in computer vision, we can anticipate robots that interact more intuitively with humans and their environment. These robots will have a heightened 'understanding' of human behaviors, facial expressions, and gestures, leading to more fluid, natural interactions. A hospital robot, for example, might be able to interpret a patient's discomfort from their facial expression, alerting the medical staff promptly.

Enhanced Autonomy: As computer vision improves, so too will the autonomy of robots. From self-driving cars to autonomous drones, these machines will become increasingly adept at navigating complex, unpredictable environments without human intervention. Enhanced object detection and recognition will be crucial in making this a reality, enabling robots to react appropriately to dynamic scenarios.

Precise Manipulation: Computer vision will empower robots with finer manipulation capabilities. For instance, in a warehouse setting, a robot could identify, locate, and handle even delicate items with precision and care, something that has been a significant challenge until now. In the realm of surgery, robots could carry out complex, intricate procedures with superhuman precision.

The journey ahead for computer vision in robotics is not without its challenges. Questions around privacy, security, and ethical use of these technologies will continue to arise and demand thoughtful solutions.  Machine learning models’ performance decays over time. Fine tuning and improving machine learning models—including computer vision ones—start with identifying where your model is not performing well. Model performance monitoring tools help to detect the outliers on which your model is failing, allowing you to address the deficiencies and significantly reduce the feedback loop. This is especially true when working with systems that have to operate in dynamic environments such as warehouses and factories. However, the promise that this field holds is remarkable. As we step into this future, we're not just crafting machines that see - we're redefining the very way we perceive and interact with the world.

Stay up to date !

Subscribe to our newsletter to get inbox notifications.
Sign up to our newsletter ↓
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.