How We Enhanced Safety Intelligence with Autonomous Emergency Braking(AEB) for Unstructured Campus Environments

Traditional AEB systems are tailored for structured urban roads with clear markings and traffic rules. Our client needed a solution purpose-built for unstructured campus terrains, where safety risks are amplified by unpredictable human and vehicle movements.

Key challenges included:

• Slow-Speed Operation – requiring more sensitive perception and braking logic.
• Irregular Road Layouts – absence of defined lanes and curbs.
• High Pedestrian Activity – unpredictable movement patterns and frequent crossings.
• Lack of Standardized Visual Cues – no signals, crosswalks, or traffic lights.

So, the main objectives were to:

• Build a MATLAB-based AEB module capable of detecting static and dynamic obstacles in low-infrastructure conditions.
• Implement time-to-collision (TTC) and safe distance-based braking logic for precise interventions.
• Ensure smooth braking control to balance passenger comfort with effective collision avoidance.
• Create a scalable MATLAB/Simulink framework that integrates seamlessly with existing ADAS modules such as Lane Keeping and Obstacle Avoidance.

Validate the AEB logic via Model-in-the-Loop (MIL) and Software-in-the-Loop (SIL) simulations, using both synthetic and real-world test cases.

Our engineering team developed a modular AEB block set using Simulink, Sensor Fusion and Tracking Toolbox, and Automated Driving Toolbox in MATLAB. This module was integrated into the EV buggy’s control pipeline to enable precise braking interventions in real time.

Before Implementation

• Vehicle lacked any automatic braking feature, relying solely on path planning and obstacle avoidance.
• Obstacle avoidance was limited to static objects and predefined waypoints.
• Pedestrian detection was inconsistent, leading to late reactions.
• No braking decision logic tied to TTC or relative velocity.

After Implementation

• Deployed a multi-sensor fusion module combining vision and radar inputs for enhanced pedestrian and vehicle detection.
• Designed a TTC-based braking algorithm that dynamically engaged brakes when collision risk exceeded safe thresholds.
• Developed a graded braking profile using PID controllers, modulating brake pressure to avoid sudden stops.
• Validated system robustness through MATLAB’s Driving Scenario Designer, simulating complex environments with dynamic pedestrian and vehicle cut-ins.

 

Key Highlights

• Multi-Sensor Fusion for Obstacle Detection:
  Combined camera and radar inputs to achieve robust obstacle detection across diverse lighting and terrain conditions.
• TTC-Based Emergency Braking Logic:
  Built a predictive braking system that engaged only when a collision was mathematically unavoidable, minimizing false activations.
• Brake Control via Simulink Controllers:
  Designed and tuned PID controllers to achieve smooth deceleration in low-speed dynamics, balancing safety with passenger comfort.
• Scenario Simulation with Driving Scenario Designer:
  Developed custom campus-specific test cases, including pedestrian crossings, sudden cut-ins, and irregular path structures.
• System-Level ADAS Integration:
  Packaged the AEB as a modular, reusable block, ensuring seamless integration with lane-keeping, obstacle avoidance, and steering modules.

• 70% Reduction in reaction time to obstacles and hazards through Real-time fusion and predictive braking logic.
• 90% Decrease in collision risk even in the absence of formal road boundaries.
• Smooth deceleration & controlled stopping power through tuned brake actuation.
• Plug-and-Play architecture to ensure reusability and scalability.