Traditional and modern approaches to robotics


Robotics, an interdisciplinary field encompassing computer vision, artificial intelligence (AI), and engineering, has seen remarkable advancements. From early experimental demonstrations like Shakey at the Stanford Research Institute (SRI) to contemporary innovations in behavior-based robotics at institutions like MIT, the evolution of robotics reflects a fusion of traditional and modern approaches. This article delves into the historical trajectory and the paradigm shift toward contemporary methodologies in robotics.

Traditional Approaches

The foundation of traditional robotics was laid with projects like Shakey, which were developed at SRI International in the late 1960s. Shakey, a pioneering mobile robot, navigated through specially designed rooms, interacting with large painted blocks and wedges to fulfill given goals. Its operation relied on a black-and-white television camera for perception, symbolic logic models for world representation, and planning algorithms like STRIPS for decision-making. However, Shakey’s success was limited to controlled environments due to its reliance on meticulously engineered surroundings.

Similarly, MIT’s endeavor with the copy-demo project demonstrated the capability of robots to replicate structures using visual perception. However, these early projects were constrained by specific environmental conditions and lacked adaptability to real-world complexities.

Computer vision aims to infer three-dimensional information from images, while AI focuses on problem-solving and planning based on symbolic representations of the world. Robotics, on the other hand, deals with physical interactions, including path planning and kinematics.

Here are the key characteristics of the traditional approach:

  • Emphasis on Internal World Models: Robots built upon a detailed internal representation of the environment, constructed through sensors and painstakingly crafted algorithms. Shakey, a groundbreaking robot at SRI, exemplified this approach by navigating pre-prepared rooms using vision and a symbolic logic model.
  • Focus on Planning and Reasoning: Actions were meticulously planned based on the internal model, with separate modules dedicated to perception, planning, and actuation. However, this sequential approach struggled with the real world’s complexities.

Modern Approach

The emergence of a new paradigm in robotics, catalyzed by dissatisfaction with traditional methodologies, began around 1984. This shift focused on reactive intelligence, real-time operation, and robust behavior in dynamic environments. Key realizations by pioneers like Agre and Chapman, Rosenschein and Kaelbling, and Brooks led to novel architectures that emphasize simplicity, modularity, and real-time interaction with the environment.

Agre and Chapman’s work at MIT, exemplified by the Pengi program, demonstrated intelligence through reactive behaviors in video game scenarios. Rosenschein and Kaelbling’s Flakey robot utilized high-level symbolic languages to generate real-time programs for goal-oriented behavior. Meanwhile, Brooks introduced the subsumption architecture, promoting a decentralized, behavior-based approach to robotics, devoid of central world models and symbolic processing.

Here are the key characteristics of the modern approach:

  • Embracing the Dynamic World: Recognizing the limitations of internal models, the modern approach prioritizes real-time adaptation. Robots like Flakey and Brooks’ creations operated directly in the environment, relying on simpler representations and reactive behaviors.
  • Behavior-Based Systems: Modular architectures like subsumption architecture replaced central planning. Smaller, interconnected modules focused on specific tasks, producing robust behavior through interaction with the world.
  • Key principles: Less reliance on explicit models, closer integration of perception and action, distributed computation, and real-time decision-making characterize this approach.

The traditional approach in robotics relied heavily on explicit world models, symbolic representations, and centralized control, limiting adaptability to dynamic environments. In contrast, modern methodologies emphasize decentralized control, reactive behaviors, and real-time interactions, facilitating robust performance in complex and uncertain scenarios.


The evolution of robotics from traditional to modern approaches reflects a paradigm shift towards simplicity, adaptability, and robustness. While traditional methodologies laid the groundwork for robotic research, contemporary approaches offer innovative solutions to the challenges of real-world deployment; by embracing the principles of simplicity, modularity, and real-time interaction, modern robotics endeavors to realize the vision of intelligent, adaptive machines capable of thriving in dynamic environments.

In summary, the journey of robotics epitomizes a convergence of diverse disciplines, from computer vision and AI to engineering, shaping a future where intelligent machines seamlessly interact with the world around them. By exploring the historical trajectory and contrasting methodologies in robotics, this article sheds light on the evolution of the field and its implications for future research and technological advancements.