Simultaneous Localisation and Mapping (SLAM)

Simultaneous Localisation and Mapping (SLAM) is a critical technology that allows a vehicle in an unfamiliar environment to estimate its exact position in real time while simultaneously constructing a map of its surroundings.

Essentially, the core purpose of SLAM is to answer two key questions at once: “Where am I?” and “What is around me?”. For instance, when an autonomous vehicle enters a basement car park where GPS reception is lost, it relies on SLAM via onboard sensors like cameras and LiDAR to “perceive” the surrounding environment—mapping out the car park layout whilst pinpointing its own location.

💡 Technology Categories

Depending on the primary sensor utilised, SLAM is generally divided into two main technological approaches, each offering distinct advantages in performance and application:

• LiDAR SLAM: Primarily relies on LiDAR sensors, capturing high-precision 3D point cloud data by measuring the reflection time of laser beams. Its key strengths are exceptional accuracy and total immunity to ambient lighting, allowing it to function reliably even in pitch-black conditions. As such, it is widely adopted in autonomous driving and robotics. However, LiDAR hardware remains relatively costly.

• Visual SLAM: Primarily relies on cameras to capture visual data, determining location and building maps by analysing image features. Its major benefits include lower hardware costs and the ability to capture rich colour and texture details, which aids advanced perception tasks like object recognition. However, its drawbacks are notable: it is highly susceptible to lighting changes, harsh shadows, and featureless surfaces (such as plain white walls), making it less stable than LiDAR SLAM.

💎 Technological Trends

In real-world autonomous driving applications, a standalone sensor setup is rarely sufficient to handle every scenario. Consequently, the prevailing industry trend is multi-sensor fusion—combining and deeply integrating data from LiDAR, cameras, Inertial Measurement Units (IMUs), and wheel speed sensors through advanced algorithms to leverage their respective strengths and offset weaknesses. For example, in GNSS-denied environments like structured underground car parks, a multi-sensor fusion approach significantly enhances the overall robustness and accuracy of localisation and mapping.