Occupancy-Driven Lighting and HVAC Setback for Parking Garages

Conditioning empty parking garages 24/7 is a significant, hidden operational drain, with lighting and ventilation accounting for up to 70% of a garage's energy use. This custom workflow automates setback by integrating vehicle detection sensors—computer vision or radar—with the Building Management System (BMS). The architecture triggers staged activation of luminaires and fans only in occupied aisles and zones, slashing energy costs and extending equipment life. Savings come from eliminating runtime on unoccupied assets, directly impacting the property's NOI.

Implementation requires integrating sensor feeds into an orchestration layer (e.g., LangGraph) that applies safety logic for minimum light levels and manages staged commands to the BMS via BACnet/IP. The workflow must include exception routing for sensor failures, audit logging for compliance, and integration with CMMS for predictive maintenance alerts based on reduced equipment cycles. Rollout is sequenced by garage zone, with performance monitored via energy dashboards against baselines.

The operational bottleneck is the continuous, wasteful operation of high-wattage luminaires and ventilation fans across vast, sporadically used garage spaces. Savings come from precise, zone-level activation triggered by vehicle detection, which reduces energy consumption and extends equipment life. The solution requires integrating computer vision or radar sensors with the Building Management System (BMS) via a central orchestration layer that executes safety logic and staged equipment commands.

Implementation connects sensor MQTT streams to an orchestration agent built on LangGraph, which maintains zone state and sequences commands. The BMS command layer uses BACnet/IP to stage fixture activation and adjust fan speeds in occupied aisles. Critical controls include fail-safe minimum light levels, exception routing for sensor faults, and rollback procedures during maintenance. Observability is built in via time-series logging of occupancy states, energy draw, and command execution for performance validation and audit.

Phase 1 establishes the foundational sensor and data layer. We deploy a pilot zone with vehicle detection sensors (radar or computer vision) and integrate them with a lightweight orchestration agent, typically built on LangGraph or a similar framework. This agent ingests raw occupancy events, logs them to a time-series database, and begins establishing baseline patterns. The goal is to validate sensor accuracy, network reliability, and data flow without touching critical BMS controls, de-risking the core integration.

Phase 2 integrates control, starting with non-critical lighting circuits. The orchestrator is connected to the BMS (e.g., Johnson Controls Metasys, Siemens Desigo) via BACnet/IP. We implement staged activation: sensors trigger lighting in occupied aisles while maintaining minimum safety levels elsewhere. Human-in-the-loop approval gates and manual override panels are installed. Phase 3 scales the validated logic to HVAC fans and dampers, adds multi-zone coordination, and deploys full observability dashboards for performance tracking and exception handling.

Governance begins with a phased rollout architecture. Initial deployment targets low-risk zones, using a canary release pattern to validate sensor accuracy and BMS command integrity before scaling. Each zone's automation logic runs in isolated containers, allowing for rollback without affecting entire garage operations. Approval gates are required for progression between phases, with success metrics tied to energy savings, false-negative occupancy detection rates, and equipment cycling counts to prevent premature wear.

Controls are embedded at each layer. The orchestration engine (e.g., LangGraph) enforces minimum light-level safety overrides, independent of occupancy signals. All BMS commands are logged with pre- and post-state values for audit. Exception handling routes sensor failures to maintenance CMMS (e.g., ServiceNow) and defaults zones to a fail-safe illuminated state. Observability dashboards track energy delta, equipment runtime, and anomaly rates, providing the data needed for confident, controlled expansion across the portfolio.