Real-Time Kinematic (RTK) GPS

The core mechanism of RTK. Unlike standard GPS that uses code-phase measurements (accurate to ~1-5 meters), RTK analyzes the carrier wave of the signal itself. The system must resolve the exact integer number of wavelengths between the satellite and receiver—the integer ambiguity. Once resolved, the phase measurement provides millimeter-level precision.

• Process: The rover receiver compares its measured carrier phase with the phase received from a known-position base station.
• Result: This differential calculation cancels out common errors (atmospheric delays, satellite clock errors), allowing the integer ambiguity to be fixed, enabling centimeter accuracy.

RTK requires a fixed base station at a known, surveyed location and one or more mobile rover units. The base station calculates error corrections by comparing its known position to its GPS-derived position and transmits these corrections to the rovers in real-time.

• Critical Link: A low-latency data link (typically UHF radio, cellular network, or Wi-Fi) is essential. Delays degrade accuracy.
• Network RTK (NRTK): Extends this concept using a network of permanent base stations, providing corrections over a wider area via an internet connection, reducing the need for a private base station.

RTK does not provide instant centimeter accuracy. It requires an initialization period called convergence time to resolve the integer ambiguities. This can take from several seconds to a few minutes, depending on satellite geometry, signal quality, and distance to the base station.

• Float Solution: The initial, less accurate position estimate (decimeter to meter-level) before integer fix.
• Fixed Solution: The high-integrity, centimeter-accurate position achieved after ambiguity resolution. The system must maintain cycle slip detection and correction to hold this fix during operation.

RTK accuracy degrades with distance from the base station due to spatial decorrelation of atmospheric errors. The ionospheric and tropospheric delays experienced by the base and rover become less similar as the distance (baseline length) increases.

• Typical Operational Range: For single-baseline RTK, optimal performance is within 10-20 km of the base station.
• Beyond this range: Integer fixing becomes more difficult, and accuracy may revert to decimeter levels. Network RTK (VRS, FKP, MAC) mitigates this by modeling atmospheric errors across a region.

Modern RTK systems utilize signals from multiple global navigation satellite systems (GNSS)—not just GPS, but also GLONASS, Galileo, and BeiDou. They also use multiple frequency bands (e.g., L1, L2, L5).

• Benefits: Increases the number of visible satellites, improving availability and reliability in challenging environments (urban canyons, near trees).
• Faster Convergence: Multi-frequency signals enable more robust and rapid resolution of the integer ambiguity, as different frequencies are affected differently by the ionosphere, allowing error modeling.

Within heterogeneous fleet orchestration, RTK GPS provides the absolute global positioning anchor for the world model. It fuses with other state estimation sources (like Visual-Inertial Odometry, LiDAR, wheel odometry) to create a unified, drift-free pose estimate for each agent.

• Sensor Fusion: RTK's absolute but sometimes intermittent signal is combined with high-frequency, relative sensors (IMU, cameras) via a Kalman Filter or similar estimator.
• Precision Requirements: Enables tight coordination (e.g., docking, precision picking, multi-robot path planning) and accurate global task assignment by providing a common, centimeter-accurate coordinate frame for all agents in the fleet.

Sensor fusion is the algorithmic process of combining data from multiple disparate sensors—such as RTK GPS, IMUs, LiDAR, and cameras—to produce a state estimate that is more accurate, consistent, and reliable than any single sensor could provide. In fleet orchestration, it is critical for robust localization.

• Purpose: Mitigates the weaknesses of individual sensors (e.g., GPS signal loss, IMU drift).
• Common Architectures: Includes Kalman filters (EKF, UKF) and optimization-based methods.
• Example: Fusing centimeter-accurate RTK position with high-frequency IMU angular rates to maintain a smooth, high-update-rate pose estimate, even during brief GPS outages.

Simultaneous Localization and Mapping (SLAM) is the computational problem where an agent builds a map of an unknown environment while simultaneously tracking its location within that map. While RTK provides absolute global position, SLAM provides relative localization and a dense environmental model.

• Complement to RTK: SLAM can operate indoors or in GPS-denied areas where RTK is unavailable.
• Sensor Inputs: Often uses LiDAR, cameras, or radar, not GNSS signals.
• Fleet Application: In a mixed warehouse, AMRs may use SLAM for fine-grained navigation between aisles, while outdoor vehicles use RTK for yard positioning.

A Kalman filter is an optimal recursive algorithm for estimating the state of a linear dynamic system from noisy measurements. The Extended Kalman Filter (EKF) adapts this for nonlinear systems, like robot motion, by linearizing around the current estimate.

• Core Mechanism: Operates in a two-step predict-update cycle, fusing a motion model with sensor observations.
• Role with RTK: The EKF is a standard workhorse for fusing RTK position fixes (which may arrive at 10-20 Hz) with high-rate IMU data (200+ Hz) to produce a smooth, continuous state estimate.
• Output: Provides not just a pose estimate, but also a covariance matrix representing the estimated uncertainty.

Visual-Inertial Odometry (VIO) is a specific sensor fusion technique that combines visual data from cameras with inertial data from an IMU to estimate a robot's 3D pose and velocity. It provides high-frequency, drift-reduced odometry without relying on external signals.

• Key Strength: Provides six-degree-of-freedom (6DOF) pose estimation, critical for drones or robots operating on uneven terrain.
• Relationship to RTK: VIO often serves as the primary odometry source, with RTK measurements used as intermittent absolute corrections to eliminate VIO's long-term drift.
• Use Case: An autonomous forklift uses VIO for stable, high-frequency control, while RTK corrections align its trajectory to a global warehouse map every second.

Dead reckoning is the process of estimating a robot's current position by advancing a previously known position based on estimates of speed, heading, and elapsed time. It is inherently prone to accumulating error, or drift.

• Primary Sensors: Typically uses wheel encoders (for ground vehicles) or an IMU.
• RTK's Role: RTK GPS is the primary method for correcting dead reckoning drift in outdoor environments. A high-accuracy RTK fix resets the accumulating error.
• System Design: In a fault-tolerant fleet, agents will dead reckon during brief RTK signal blockages (e.g., under a canopy), using the last known good fix and motion model until a correction is reacquired.