State of Charge (SoC)

State of Charge (SoC) is a measurement, expressed as a percentage, indicating the current available energy capacity of a battery relative to its fully charged state. It is the primary indicator of an agent's immediate operational endurance.

• Key Calculation: SoC (%) = (Remaining Capacity / Total Usable Capacity) * 100.
• Not to be confused with State of Health (SoH), which measures the battery's overall degradation and maximum capacity loss over its lifetime.
• In fleet orchestration, SoC is a critical input for battery-aware scheduling and predictive maintenance systems.

SoC cannot be measured directly; it must be estimated through a combination of methods, each with trade-offs between accuracy, cost, and complexity.

• Coulomb Counting (Bookkeeping): Integrates current flow in and out of the battery over time. It's simple but prone to error drift due to measurement inaccuracies and unknown self-discharge.
• Voltage-Based Estimation: Uses the battery's open-circuit voltage (OCV), which has a known relationship to SoC. Accurate only when the battery is at rest, as load causes voltage sag.
• Model-Based Estimation (Kalman Filters): Combines coulomb counting with voltage and temperature readings using a mathematical model of the battery's electrochemical behavior. This is the most accurate method for dynamic, in-operation estimation used in advanced systems.
• Impedance Spectroscopy: Measures the battery's internal resistance and reactance, which change with SoC. Highly accurate but computationally intensive.

SoC data directly drives several critical autonomous decision-making loops within a fleet management platform.

• Dynamic Task Allocation: Agents with low SoC are assigned shorter-duration or lower-priority tasks near charging stations.
• Spatial-Temporal Scheduling: The orchestration engine plans charging cycles as explicit tasks, scheduling agents to docks based on predicted energy consumption for their assigned routes.
• Graceful Degradation: A robot may enter a low-power mode or limit its speed as SoC drops, ensuring it can reach a charger rather than stranding itself.
• Load Balancing: Workload is distributed to prevent a subset of agents from depleting simultaneously, creating charging bottlenecks.

SoC interacts with State of Health (SoH) and battery degradation in ways that affect long-term fleet planning.

• Degradation Acceleration: Consistently operating at very high (e.g., >90%) or very low (e.g., <20%) SoC accelerates chemical aging, reducing the battery's total usable capacity (SoH).
• Remaining Useful Life (RUL) Forecasting: SoC cycling data (depth and frequency of discharge) is a key input for predictive models that estimate a battery's RUL.
• Calibration: Periodic full charge/discharge cycles are often required to recalibrate SoC estimation algorithms and maintain accuracy as the battery degrades.

SoC is a core component of the telemetry stream from each agent, feeding into the fleet-wide view and metrics pipeline.

• Golden Signal for Power: SoC is a primary metric for system Saturation (resource exhaustion) within the SRE framework.
• Anomaly Detection: A rapidly falling SoC that deviates from the energy consumption model can indicate a mechanical fault (e.g., stuck brake, bearing failure) or a failing battery cell.
• Health Score Integration: SoC is a weighted input into a composite health score, triggering alerts when thresholds are breached (e.g., SoC < 15%).
• Data is used to enforce Service Level Objectives (SLOs) related to fleet availability and task completion rates.

Defined SoC thresholds trigger automated actions to protect hardware and ensure safety.

• Critical Low (e.g., 5-10%): Forces an immediate stop and safe shutdown. The agent broadcasts a high-priority assistance request.
• Low (e.g., 15-20%): Triggers a return-to-charger command. The real-time replanning engine finds the optimal path to the nearest available dock.
• High (e.g., 80-90%): For lithium-based batteries, charging often slows down (constant-voltage phase) above ~80% to protect longevity. Scheduling systems may treat an agent as 'available' once it passes this threshold.
• Safety Buffer: Operations are planned within a usable SoC window (e.g., 20%-80%) to maximize battery life and always maintain a reserve for emergency maneuvers or replanning.

The irreversible reduction in a battery's maximum capacity and performance over time due to chemical aging. Key factors include:

• Charge cycles: Each full discharge/charge cycle gradually reduces capacity.
• Temperature exposure: High temperatures accelerate chemical breakdown.
• Depth of Discharge (DoD): Regularly draining a battery deeply increases degradation rate. This process directly impacts the State of Charge (SoC) calculation, as the 'fully charged' reference point diminishes. Monitoring degradation is essential for accurate Remaining Useful Life (RUL) predictions and battery-aware scheduling.

A class of optimization algorithms for task and route planning that incorporates the energy constraints and charging cycles of agents. This involves:

• Dynamically assigning tasks based on current State of Charge (SoC) and required energy expenditure.
• Planning optimal times for agents to visit charging stations to minimize total fleet downtime.
• Balancing workload to prevent some agents from degrading faster than others. Effective implementation requires integration with fleet state estimation and real-time replanning engines to adapt to changing energy levels.

A forecast metric estimating the remaining operational time or cycles before a battery or component is expected to fail. RUL is a core output of predictive maintenance systems and is calculated using:

• Historical telemetry streams of charge/discharge patterns.
• Models of battery degradation rates.
• Real-time measurements like internal resistance and temperature. Unlike State of Charge (SoC), which measures current energy, RUL predicts future longevity. Accurate RUL allows for proactive part replacement, minimizing unplanned agent downtime.

A maintenance strategy that uses data analysis and machine learning to predict equipment failures before they occur, scheduling repairs during planned downtime. For battery systems, this relies on:

• Continuously monitoring State of Charge (SoC), voltage, and temperature telemetry streams.
• Detecting anomalies in charge/discharge curves that signal impending failure.
• Calculating Remaining Useful Life (RUL) forecasts. This approach contrasts with reactive (fix-after-failure) or periodic (time-based) maintenance, optimizing fleet availability and reducing costs.

A continuous, real-time flow of operational data from agents to a central collection system. For battery and health monitoring, key data points in this stream include:

• State of Charge (SoC) percentage.
• Voltage, current, and temperature readings.
• Health check API responses and heartbeat signals. This raw data feeds into metrics pipelines for aggregation, enables anomaly detection, and provides the historical dataset required for training predictive maintenance models.

A composite numerical value that summarizes the overall operational status of an agent, derived from multiple underlying metrics. A comprehensive health score for a mobile robot often incorporates:

• State of Charge (SoC) and battery voltage stability.
• Responsiveness from liveness and readiness probes.
• Diagnostic codes from self-diagnostics.
• Performance metrics like compute load or communication latency. This synthesized score provides a quick, at-a-glance fleet-wide view for site managers, triggering alerts when it falls below a defined threshold.