A Feasibility-Enhanced Control Barrier Function Method for Multi-UAV Collision Avoidance

Qishen Zhong; Junlong Wu; Jian Yang; Guanwei Xiao; Junqi Wu; Zimeng Jiang; Pingan Fang

cs.ROMar 13, 2026

A Feasibility-Enhanced Control Barrier Function Method for Multi-UAV Collision Avoidance

Authors: Qishen Zhong, Junlong Wu, Jian Yang, Guanwei Xiao, Junqi Wu, Zimeng Jiang, Pingan Fang

Organizations: School of Automation Science and Engineering, South China University of Technology, Guangzhou 510641, Guangdong, China

Abstract

This paper presents a feasibility-enhanced control barrier function (FECBF) framework for multi-UAV collision avoidance. In dense multi-UAV scenarios, the feasibility of the CBF quadratic program (CBF-QP) can be compromised due to internal incompatibility among multiple CBF constraints. To address this issue, we analyze the internal compatibility of CBF constraints and derive a sufficient condition for internal compatibility. Based on this condition, a sign-consistency constraint is introduced to mitigate internal incompatibility. The proposed constraint is incorporated into a decentralized CBF-QP formulation using worst-case estimates and slack variables. Simulation results demonstrate that the proposed method significantly reduces infeasibility and improves collision avoidance performance compared with existing baselines in dense scenarios. Additional simulations under varying time delays demonstrate the robustness of the proposed method. Real-world experiments validate the practical applicability of the proposed method.

Figures & tables

Fig. 1: Illustration of internal compatibility and incompatibility among multiple CBF constraints in multi-UAV collision avoidance. Compatibility refers to the existence of at least one control input that simultaneously satisfies all CBF constraints within the admissible input bounds. Internal compatibility specifically denotes the mutual consistency of the CBF constraints themselves, meaning that their intersection is non-empty without considering the admissible input bounds. Each ellipse represents a CBF constraint, while the circle denotes the admissible input bounds.

Fig. 2: The cones of the sign-consistency constraint ( 21 ) . $\beta_{1}<\beta_{2}<\beta_{3}$ are the half-apex angles of these cones, respectively.

Fig. 3: Overview of the proposed FECBF method.

Para.	Val.	Para.	Val.	Para.	Val.
$\theta^{\max}_{i}$	${\pi}/{2}$	$v^{\max}_{i}$	$[2,3]\,\mathrm{m/s}$	$\gamma^{\max}_{i}$	$\pi/36$
$\theta^{\min}_{i}$	$-{\pi}/{2}$	$v^{\min}_{i}$	${v^{\max}_{i}}/{4}$	$\gamma^{\min}_{i}$	$-\pi/36$
$\psi^{\max}_{i}$	$2{\pi}$	$a^{\max}_{i}$	$1\,\mathrm{m/s}$	$\omega^{\max}_{i}$	$\pi/18$
$\psi^{\min}_{i}$	$0$	$a^{\min}_{i}$	$-1\,\mathrm{m/s}$	$\omega^{\min}_{i}$	$-\pi/18$
$\zeta$	$0.5$	$\kappa$	$0.08$	$\beta$	$7\pi/24$
$\lambda$	$3$

TABLE I: Key experimental parameters used in simulations. Angular quantities are given in radians.

Panel of Figure 5 — Fig. 4: Illustration of three representative and challenging simulation scenarios used for performance evaluation. The gray shaded region denotes the ground plane, while the points with halo and arrows represent the positions and velocity vectors of the UAVs, respectively. (a) Convergence scenario; (b) Dual-circle scenario, depicted from a top view; (c) Head-on scenario.

Scenario	$n$	SR [%] $\uparrow$			IC $\downarrow$			AT [s] $\downarrow$			CT [ms] $\downarrow$
		Ours	VOCBF	DRCBF	Ours	VOCBF	DRCBF	Ours	VOCBF	DRCBF	Ours	VOCBF	DRCBF
	50	97.98	96.72	95.92	96.58	114.50	121.11	377.28	382.14	384.70	1.10	1.14	0.96
Convengence	100	96.72	95.56	93.66	147.73	182.82	184.41	408.32	419.69	418.65	1.32	1.61	1.08
	150	95.89	94.01	91.59	193.51	240.49	242.01	435.45	448.91	447.32	1.50	1.89	1.21
	50	100.00	93.02	93.16	75.31	141.44	139.78	332.99	337.39	336.86	0.86	0.80	0.62
Dual-circle	100	100.00	93.56	93.74	74.59	141.45	139.79	332.97	337.38	336.85	1.10	1.05	0.96

TABLE II: Performance Comparison of Different Methods Across Scenarios

Fig. 5: Time delay simulation results for all three scenarios with

n=150

. The bottom-right subfigure shows the results for the dual-circle scenario with

\tau

ranging from

3

s to

5

s at intervals of

0.2

s, while the remaining subfigures show results for

\tau\in\{1,3,5\}

Fig. 6: Snapshots of real-world experiments with four, six, and eight UAVs. Each row corresponds to a different UAV counts, while columns represent the initial configuration, the moment of closest approach during the collision avoidance, and the post-avoidance configuration.

Explore similar work

A Temporal Barrier Framework for Collision Avoidance in Multi-Agent Autonomous Aerial Vehicles
Aug 14, 2026Benedikt Barthel Sorensen, Mitchell Black, Erfaun Noorani +1Collision Avoidance Aerial Robotics
Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA · MIT Lincoln Laboratory, Lexington, MA, USA
Exact Feasibility Certification and Optimal Responsibility Allocation for Multi-Robot CBF Safety Filters
Sep 14, 2026Chandan Kumar Sah, Jishnu KeshavanSafety-Critical Control Control Barrier Functions
Department of Mechanical Engineering, Indian Institute of Science, Bengaluru 560012, India
RVC-NMPC: Nonlinear Model Predictive Control with Reciprocal Velocity Constraints for Mutual Collision Avoidance in Agile UAV Flight
Dec 9, 2025Vit Kratky, Robert Penicka, Parakh M. Gupta +2Multi-Robot Motion Planning Nonlinear MPC
Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague, Czech Republic

Aug 14, 2026J/K move · Enter open · S save

A Temporal Barrier Framework for Collision Avoidance in Multi-Agent Autonomous Aerial Vehicles

Benedikt Barthel Sorensen, Mitchell Black, Erfaun Noorani, Themistoklis Sapsis

Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA · MIT Lincoln Laboratory, Lexington, MA, USA

Collision Avoidance Aerial Robotics Control Barrier Functions

Open paper AskPDF arXiv

Operating teams of autonomous aircraft in dynamic, uncertain, and potentially adversarial environments requires safety protocols that are reliable yet selective, and allow agents to fly in close proximity while making progress toward mission objectives. We introduce adversarial time-to-collision (aTTC), a risk metric that quantifies, for a given agent, how quickly any surrounding agent could reach it assuming adversarial intent. We embed aTTC into the control barrier function (CBF) framework, defining the barrier directly in time rather than distance or velocity. The resulting aTTC-CBF is inherently anticipatory: agents modulate their own velocity based not on whether a peer is on a collision course, but on how quickly one could reach collision given its dynamical constraints. A differentiable neural-network surrogate makes the aTTC computable in real time within a standard CBF quadratic program. Across long time-horizon simulations of 3D independent-pursuit and formation-flight scenarios, the aTTC-CBF achieves up to twice the waypoint progress at half the collision rate of a higher-order distance-based CBF baseline.

Abstract

Explore similar work

A Temporal Barrier Framework for Collision Avoidance in Multi-Agent Autonomous Aerial Vehicles

Exact Feasibility Certification and Optimal Responsibility Allocation for Multi-Robot CBF Safety Filters

RVC-NMPC: Nonlinear Model Predictive Control with Reciprocal Velocity Constraints for Mutual Collision Avoidance in Agile UAV Flight