Paper deep dive
Decentralized Opinion-Integrated Decision making at Unsignalized Intersections via Signed Networks
Bhaskar Varma, Ying Shuai Quan, Karl D. von Ellenrieder, Paolo Falcone
Intelligence
Status: succeeded | Model: google/gemini-3.1-flash-lite-preview | Prompt: intel-v1 | Confidence: 94%
Last extracted: 4/14/2026, 1:38:01 AM
Summary
The paper proposes a decentralized decision-making framework for connected autonomous vehicles (CAVs) at unsignalized intersections using dual signed networks: a conflict-based communication network and a commitment-driven belief network. By integrating opinion dynamics with a closed-form predictive feasibility gate, the system enables vehicles to reach collision-free crossing commitments (GO/YIELD) without a centralized coordinator, outperforming traditional First-Come-First-Served (FCFS) policies in complex traffic scenarios.
Entities (5)
Relation Signals (3)
Signed Networks → enables → Decentralized Coordination
confidence 95% · enable cooperation without a centralized coordinator
Connected Autonomous Vehicles → uses → Signed Networks
confidence 95% · vehicles exchange intent through dual signed networks
Predictive Feasibility Gate → determines → Crossing Order
confidence 90% · Crossing order emerges from geometric feasibility and arrival priority
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:In this letter, we consider the problem of decentralized decision making among connected autonomous vehicles at unsignalized intersections, where existing centralized approaches do not scale gracefully under mixed maneuver intentions and coordinator failure. We propose a closed-loop opinion-dynamic decision model for intersection coordination, where vehicles exchange intent through dual signed networks: a conflict topology based communication network and a commitment-driven belief network that enable cooperation without a centralized coordinator. Continuous opinion states modulate velocity optimizer weights prior to commitment; a closed-form predictive feasibility gate then freezes each vehicle's decision into a GO or YIELD commitment, which propagates back through the belief network to pre-condition neighbor behavior ahead of physical conflicts. Crossing order emerges from geometric feasibility and arrival priority without the use of joint optimization or a solver. The approach is validated across three scenarios spanning fully competitive, merge, and mixed conflict topologies. The results demonstrate collision-free coordination and lower last-vehicle exit times compared to first come first served (FCFS) in all conflict non-trivial configurations.
Tags
Links
- Source: https://arxiv.org/abs/2604.09351v1
- Canonical: https://arxiv.org/abs/2604.09351v1
Trouble viewing inline? Open PDF directly →
Full Text
27,625 characters extracted from source content.
Expand or collapse full text
Decentralized Opinion-Integrated Decision making at Unsignalized Intersections via Signed Networks Bhaskar Varma∗, Ying Shuai Quan∗, Karl D. von Ellenrieder∗ and Paolo Falcone∗ ∗Bhaskar Varma and Karl D. von Ellenrieder are with the Faculty of Engineering, Free University of Bozen-Bolzano, Italy sbalagopala, karl.vonellenrieder@unibz.it. ∗Ying Shuai Quan and Paolo Falcone are with the Mechatronics Group, Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden quany,paolo.falcone@chalmers.se. Abstract In this letter, we consider the problem of decentralized decision making among connected autonomous vehicles at unsignalized intersections, where existing centralized approaches do not scale gracefully under mixed maneuver intentions and coordinator failure. We propose a closed-loop opinion-dynamic decision model for intersection coordination, where vehicles exchange intent through dual signed networks: a conflict topology based communication network and a commitment-driven belief network, enable cooperation without a centralized coordinator. Continuous opinion states modulate velocity optimizer weights prior to commitment; a closed-form predictive feasibility gate then freezes each vehicle’s decision into a go or yield commitment, which propagates back through the belief network to pre-condition neighbor behavior ahead of physical conflicts. Crossing order emerges from geometric feasibility and arrival priority without the use of joint optimization or a solver. The approach is validated across three scenarios spanning fully competitive, merge, and mixed conflict topologies. The results demonstrate collision-free coordination and lower last-vehicle exit times compared to first come first served (FCFS) in all conflict-non-trivial configurations. I Introduction and Related Work Urban unsignalized intersections represent one of the most demanding coordination environments for connected autonomous vehicles. While the problem is not new and looks solved using a centralized model predictive control (MPC) like [11], the main issue is that it relies on a centralized coordinator and considers only longitudinal control along straight paths, a limitation that persists even in decentralized MPC extensions [6]. Reservation-based methods [13] treat the intersection as a shared resource and assign time slots through a centralized intersection manager. Some intersection managers propose [7] a First Come First Served (FCFS) policy instead of an optimal passing order; this works adequately if you only consider straight-path maneuvers, but with turn maneuvers the overall throughput will be compromised. Recent approaches take the lane geometry and turns into consideration [8], but again reintroduce a centralized controller, which increases the computational complexity and dependence when upscaled. Also, learning-based optimization algorithms have been proposed [5] yet remain dependent on offline training and generalize poorly across unseen configurations. These approaches either require a persistent central authority that fails to scale gracefully under coordinator failure or intermittent communication, or rely on optimization-based joint and distributed MPC formulations that achieve strong performance at the cost of computational complexity. On the other hand, opinion dynamics [1] have recently attracted attention especially for decentralized, interactive decision making: including robots interacting with human movers in [4] and for task allocation in uncrewed surface vessels [10], and have also been used in unsignalized roundabouts [9]. Opinion dynamics on signed social networks [12] offer a different paradigm: distributed agents form opinions through local interactions, and bifurcation phenomena naturally produce differentiated equilibria without explicit negotiation. The natural bifurcation of signed-network equilibria into two opposing committed states maps directly onto the binary GO/YIELD decision required at intersections. The multitopic extension in [2] establishes conditions under which structurally balanced signed networks produce stable committed equilibria, providing theoretical grounding for opinion-based coordination. The primary focus of this work is to study interactive decision making at intersections, not throughput maximization as in high-density traffic management. While extensive literature addresses the latter, the core problem of individual vehicle decision-making is often bypassed through centralized sequencing, rather than vehicles resolving it distributively. We address this gap with a decentralized architecture integrating opinion dynamics and decomposed predictive control, where a closed-form feasibility gate freezes each vehicle’s commitment. Opinion evolution over dual signed networks facilitates continuous coordination without a central authority. This letter makes the following contributions: 1. Dual signed graph architecture. A conflict based communication graph and commitment driven belief network together define structured Suppression, Permission, and Coordination influence, routing committed state through opinion dynamics without a weighted coupling matrix. 2. Decomposed predictive coordination. MPC is decomposed into a closed-form predictive feasibility check, which drives a decentralized commitment gate and a single-step multi-objective velocity optimizer, while preserving predictive safety. 3. Opinion-integrated control. The continuous opinion modulates optimizer weights before commitment; each commitment event updates the belief network, triggering opinion spikes that propagate intent through the network ahead of physical conflict resolution. 1W-IN8W-OUT5E-IN4E-OUT3S-IN2S-OUT7N-IN6N-OUTRevoR_evoRintR_int1→ 21→ 41→ 6 Figure 1: Four-way unsignalized intersection. In-lanes (blue, labeled 1,3,5,7) and out-lanes (red, labeled 2,4,6,8). Maneuvers from lane 1: right turn 1→ 2 (green), straight 1→ 4 (orange), left turn 1→ 6 (gray). Red circle: evolution zone Revo=15R_evo=15 m I Problem Statement I-A Intersection Geometry We consider a four-way unsignalized intersection with right-hand traffic rules as shown in Fig. 1. Each approach (West, South, East, North) has two lanes: an in-lane for approaching vehicles and an out-lane for departing vehicles. In-lanes are labeled 1,3,5,71,3,5,7 (West, South, East, North) and out-lanes 2,4,6,82,4,6,8 respectively. I-B Problem Statement Given a set of CAVs approaching a four-way unsignalized intersection with mixed maneuver intentions, design a distributed coordination system such that: (1) Safety: all vehicles complete their intended maneuvers without collision; (2) Decentralization: without central coordinator or pre-assigned crossing order. I-C Vehicle Model and Path Execution Consider NaN_a CAVs with state i=[xi,yi,θi,vi]⊤x_i=[x_i,y_i, _i,v_i] . Each vehicle follows a pre-computed arc-length parameterized path determined by its declared (in-lane, out-lane) intention. Path tracking is assumed ideal: lateral position is read directly from the path. The coordination layer determines v˙i=ai(t) v_i=a_i(t); path geometry serves as a fixed spatial reference along which each vehicle progresses.The length of each vehicle LvL_v is 4.5m and the width is 1.8m for the simulations. I Method I-A Signed Opinion Dynamics Model We build on the nonlinear opinion dynamics framework of [2]. The NaN_a agents are on a signed network with adjacency matrix A∈ℝNa×NaA ^N_a× N_a, where aik>0a_ik>0 denotes cooperative and aik<0a_ik<0 antagonistic interactions between agents i and k. Let zij∈ℝz_ij be the opinion of agent i about option j. The continuous-time dynamics are τz˙ij=−dzij+tanh(uIij+bij),τ\, z_ij=-d\,z_ij+ \! (u\,I_ij+b_ij ), (1) where the internal state is Iij= I_ij=\; αzij+γ∑k≠iaikzkj α\,z_ij+γ _k≠ ia_ik\,z_kj +β∑l≠jgjlzil+δ∑k≠i∑l≠jaikgjlzkl. +β _l≠ jg_jl\,z_il+δ _k≠ i _l≠ ja_ik\,g_jl\,z_kl. (2) Here d>0d>0 is resistance, u≥0u≥0 is attention, bijb_ij is external bias, and gjlg_jl are entries of the belief-system graph G. The four governing parameters are: α (self-reinforcement), γ (inter-agent coupling via signed A), β (cross-option coupling via G), and δ (combined inter-agent and cross-option). The neutral state Z=0Z=0 undergoes a pitchfork bifurcation at a critical attention u∗u^*. Above u∗u^* agents rapidly commit to non-neutral opinions, a property we exploit for vehicle coordination via zone-based attention. I-B Zone-Based Attention and Commitment Mechanism The intersection region is partitioned into three zones according to each vehicle’s distance from the intersection center (which is the origin), as shown in Fig. 1, where di(t)=‖i(t)‖d_i(t)=\|p_i(t)\| and zonei(t)=EVOLUTIONdi<Revo,Revo=15m,DECISIONdi≤Rint,Rint=5m,EXITEDdi>Rint.zone_i(t)= casesEVOLUTION&d_i<R_evo,\;R_evo=15\,m,\\ DECISION&d_i≤ R_int,\;R_int=5\,m,\\ EXITED&d_i>R_int. cases Opinion dynamics are zone-aware through a state-dependent attention gain ui(t)=u0(zonei)+Ku(zi−12)2,u_i(t)=u_0\! (zone_i )+K_u\! (z_i- 12 )^\!2, (3) with u0=0.5u_0=0.5 (EVOLUTION), u0=0.8u_0=0.8 (DECISION), and Ku=2.0K_u=2.0. As ziz_i departs from neutral, uiu_i increases and accelerates opinion formation analogous to the supercritical bifurcation in (1). Each vehicle maintains a discrete commitment state σi∈N,G,Y,E, _i∈ \N,\;G,\;Y,\;E \, (4) where the letters denote negotiate, go, yield, and exit. Initially σi=N _i=N. Opinion evolves continuously while σi=N _i=N across both the evolution and decision zones, with attention gain uiu_i increasing as the vehicle approaches the intersection. Upon entering the decision zone, the commitment gate evaluates FCFS priority and feasibility, freezing opinion at commitment: it freezes at zi→1z_i→1 on G and remains frozen through exit, while for zi→0z_i→0 on Y, the vehicle only resumes negotiation once the conflicting vehicle exits. This commitment mechanism facilitates cooperative collision free passing, which is discussed in more detail in Section I-F. I-C Conflict Topology Based Communication Network Graph For each vehicle pair (i,j)(i,j) we compute a crossing conflict indicator K(i,j)=1paths geometrically cross ,0otherwise,K(i,j)= cases1&paths geometrically cross ,\\ 0&otherwise, cases and a merge indicator M(i,j)=1out-lanei=out-lanej,0otherwise,M(i,j)= cases1&out-lane_i=out-lane_j,\\ 0&otherwise, cases showing whether they share a common out-lane. Both crossing and merge conflicts require sequencing; M is maintained separately to distinguish the type of deferral in the trajectory optimization layer. We define the static signed adjacency matrix A∈−1,0,+1Na×NaA∈\-1,0,+1\^N_a× N_a, where A(i,j)=−1K(i,j)=1 or M(i,j)=1,+1K(i,j)=0 and M(i,j)=0,0i=j.A(i,j)= cases-1&K(i,j)=1 or M(i,j)=1,\\ +1&K(i,j)=0 and M(i,j)=0,\\ 0&i=j. cases (5) When A(i,j)=−1A(i,j)=-1 we have an antagonistic edge (conflict requires sequencing) and when A(i,j)=+1A(i,j)=+1 we have a cooperative edge (no conflict). Since geometric conflict is symmetric, A(i,j)=A(j,i)A(i,j)=A(j,i) and A is generally an undirected signed graph, calculated using declared maneuver intentions. I-D Commitment State Aware Belief System Network Graph The dynamic belief matrix (t)∈−1,0,+1Na×NaB(t)∈\-1,0,+1\^N_a× N_a encodes the current commitment state of each neighbor, received via V2V B(i,j)=+1σj=go,−1σj=yield,0otherwise.B(i,j)= cases+1& _j= go,\\ -1& _j= yield,\\ 0&otherwise. cases (6) Unlike A, the belief matrix is directed B(i,j)≠B(j,i)B(i,j)≠ B(j,i), since σi _i and σj _j evolve independently, and is sparse throughout most of the scenario. It is nonzero only when a neighbor has committed and is actively in the decision or intersection zone. An uncommitted neighbor (B(i,j)=0B(i,j)=0) exerts no influence on i’s opinion, preventing undecided vehicles from disrupting each other’s opinion trajectories. I-E Proposed Decision Model for Intersection Each vehicle i maintains a scalar opinion zi∈[0,1]z_i∈[0,1], where zi→1z_i→ 1 represents GO, zi→0z_i→ 0 YIELD , and zi=0.5z_i=0.5 neutral. Each committed neighbor j (i.e. B(i,j)≠0B(i,j)≠ 0) is assigned to one of the three interaction channels based on the joint state of A(i,j)A(i,j) and B(i,j)B(i,j) +−(i)=j:A(i,j)=−1,B(i,j)=+1, ^-_+(i)=\j:A(i,j)=-1,\;B(i,j)=+1\, −(i)=j:A(i,j)=−1,B(i,j)=−1, ^-_-(i)=\j:A(i,j)=-1,\;B(i,j)=-1\, +(i)=j:A(i,j)=+1,B(i,j)≠0. ^+(i)=\j:A(i,j)=+1,\;B(i,j)≠0\. This is our key departure from (2), where aika_ik is a continuous multiplier. Here A and B determine which channel a neighbor belongs to, while independent channel gains govern influence strength. The three channel signals are: Suppression i=maxj∈+−(i)zj,S_i= _j\,∈\,N^-_+(i)z_j, where conflicting neighbor-committed GO drives zi→0z_i→ 0. Permission i=meanj∈−(i)(1−zj),P_i=mean_j\,∈\,N^-_-(i)(1-z_j), where conflicting neighbor-commited YIELD drives zi→1z_i→ 1. Coordination i=meanj∈+(i)zj,C_i=mean_j\,∈\,N^+(i)z_j, where cooperative neighbor- committed GO/YIELD drives alignment. The internal opinion state aggregating all three channels is Ii=αs(zi−12)−αSi+αPi+αCi,I_i= _s\! (z_i- 12 )- _S\,S_i+ _P\,P_i+ _C\,C_i, (7) where αS>αP>αC _S> _P> _C ensures suppression always dominates for safety. The opinion-based decision dynamics in (1) are adopted from the range [−1,1][-1,1] to the range [0,1][0,1] so τzz˙i=−dzi+12(1+tanh(uiIi)), _z\, z_i=-d\,z_i+ 12\! (1+ (u_i\,I_i) ), (8) where d=1d=1 and τz=0.1 _z=0.1 s. The (1+tanh(x))/2(1+ (x))/2 maps IiI_i to (0,1)(0,1) without boundary saturation, and the neutral point zi=0.5z_i=0.5 provides symmetric recovery after resumption. The opinion-like dynamics in (8) are inspired by signed-graph nonlinear opinion dynamics but the commitment decision is determined entirely by the predictive gate, not by any opinion threshold crossing. I-F Predictive Commitment Gate and Velocity Optimization Rather than applying MPC as a monolithic trajectory optimizer, we decompose its two constituent ideas: a) predictive constraint evaluation drives the commitment gate, determining safe crossing order; and b) single-step cost minimization drives velocity regulation, ensuring smooth speed adaptation consistent with σi _i, as explained by (4). This decomposition eliminates the need for a QP solver while preserving the safety guarantees of predictive constraint satisfaction. On entry to the decision zone, vehicle i records its arrival timestamp tidzt_i^dz and broadcasts it via V2V communication as shown in Fig. 2 alongside the states (σi,zi,i)( _i,z_i,x_i). Predictive feasibility check: Vehicle i projects its earliest crossing window, Ti=[t+divimax,t+di+Lboxvimax]T_i= [t+ d_iv _i,\;\;t+ d_i+L_boxv _i ] where Lbox=2Rint+LvL_box=2R_int+L_v. Feasibility requires non-overlapping occupancy windows against all σj=G _j=G neighbors with K(i,j)=1K(i,j)=1 min(Ti(2),Tj(2))−max(Ti(1),Tj(1))<ε,ε=0.6s. \! (T_i(2),T_j(2) )- \! (T_i(1),T_j(1) )< , =0.6\,s. (9) Feasible implies σi→G _i→ G, TiT_i stored and broadcast; infeasible implies σi→Y _i→ Y. For merge conflicts M(i,j)=1M(i,j)=1, a time-to-merge-point check with margin tmargin=1.5t_margin=1.5 s replaces (9). Each commitment event updates B(j,i)B(j,i) for all neighbors via V2V, activating the channels in (8) and inducing an opinion spike among required vehicles, either zj→0z_j→0 or zj→1z_j→1 mimicking [3]. Single-Step Multi-Objective Velocity Optimization Each vehicle follows its pre-computed geometric path exactly via arc-length parameterization. The Optimization layer determines only longitudinal acceleration ai∗a_i^* at each timestep solving a multi objective cost function modulated by opinion, minimizea∈Ji(a)=Jprog+Jcomf+Jspat+Jyield [ $ a subject~to$][l] a minimize J_i(a)=J_prog+J_comf+J_spat+J_yield subjectto [ $ a subject~to$][c]subject~to a a ∈[−5.0, 2.5]m/s2 ∈[-5.0,\;2.5]\;m/s^2 where A is a uniform grid of Nc=15N_c=15 candidates over the acceleration limits. And vimax∈11.1, 8.0, 7.0v _i∈\11.1,\,8.0,\,7.0\ m/s is the speed limit for straight, left and right maneuvers respectively, ensuring safe cornering speeds across all maneuver types. a) Progress Cost: Jprog=wp(σ,z)dimax(vpred, 0.1).J_prog=w_p^(σ,z)\, d_i (v_pred,\,0.1). The predicted velocity is clipped to a maneuver-dependent speed limit vpred=max(0,min(vi+aΔt,vimax)),v_pred=max(0,min(v_i+a t,v _i)), vimaxv_i reflects the geometric constraint of the assigned maneuver and wp(σ,z)=10wpσi=G,di≤Rint,(0.5+zi)wpσi=N,wpotherwise.w_p^(σ,z)= cases10\,w_p& _i=G,\;d_i≤ R_int,\\ (0.5+z_i)\,w_p& _i=N,\\ w_p&otherwise. cases In the negotiating state, zi→1z_i→ 1 (GO) increases approach aggressiveness while zi→0z_i→ 0 (YIELD) reduces it, providing a smooth pre-commitment speed adaptation. Where wpw_p = 1. b) Comfort Cost: Acceleration effort is penalized quadratically to suppress aggressive maneuvers Jcomf=wca2,J_comf=w_c\,a^2, where wc=0.5w_c=0.5. c) Spatial Repulsion Cost: from conflicting neighbors K(i,j)=1K(i,j)=1 Jspat=∑j∈Kiwjexp(−12(dij−dsafe)),J_spat=\! _j\,∈\,K_iw_j\; \! (- 12(d_ij-d_safe) ), wj=wcomσj=G,wdec(0.5+zj)σj=N,dj<Revo,wevootherwise.w_j= casesw_com& _j=G,\\ w_dec\,(0.5+z_j)& _j=N,\;d_j<R_evo,\\ w_evo&otherwise. cases A negotiating neighbor leaning GO (zj→1z_j→1) induces stronger repulsion than one leaning YIELD (zj→0z_j→0), reflecting its expressed crossing intent before commitment. JspatJ_spat is suppressed for crossing conflicts when σi=G _i=G and di≤Rintd_i≤ R_int, where wcomw_com = 1000, wdecw_dec = 10, wevow_evo = 1 and the safe distance dsafed_safe = 3m. d) Yield Braking Cost: When σi=Y _i=Y, a braking cost is applied with exponential onset beyond RintR_int\, ensuring smooth deceleration, Jyield=wcomvpreddi≤Rint,0.05wcomvprede−(di−Rint)/6di>Rint,J_yield= casesw_com\,v_pred&d_i≤ R_int,\\[2.0pt] 0.05\,w_com\,v_pred\,e^-(d_i-R_int)/6&d_i>R_int, cases avoiding discontinuous braking at the stop boundary. V2V BROADCASTNETWORK AAStatic ⋅· SignedConflict topologyNETWORK BBDynamic ⋅· DirectedCommitment-drivenOPINION DYNAMICSSuppressionYield intent PermissionGo intent CoordinationSoft consensus PREDICTIONFCFSPriority order FeasibilityGo / Yield VELOCITY OPTIMIZEROpinion-modulated cost functionProgress ⋅· Comfort ⋅· Yield Figure 2: Architecture of the proposed framework IV Simulation Results and Analysis We validate the proposed framework in MATLAB across all 81 combinatorial maneuver scenarios involving 4 CAVs approaching a 4-way unsignalized intersection, achieving improved performance over FCFS across all configurations. While the all-right-turn case gives a fully cooperative signed network with trivially parallel passing and no sequencing challenge, 3 representative cases are presented as in Table I. Scenario 1 introduces maximum contention: all 4 vehicles intend left turns, every pairwise edge in A is competitive (-1), Vehicles resolve priority through FCFS ordering based on decision zone arrival time, with opinions evolving accordingly under full suppression channel activation. As shown in Fig. 4 and the opinion and speed profiles in Fig. 3, CAV1 is granted priority first, followed by CAV3, then CAV2, and finally CAV4, each commitment propagating through B and pre-conditioning the subsequent vehicle’s opinion transition. All vehicles traverse the intersection conflict-free, demonstrating that the signed network structure resolves fully competitive topologies without any explicit sequencing constraint. Scenario 2 introduces complex conflict topology, comprising two merge conflicts and one crossing conflict under mixed maneuver intentions. FCFS priority is respected as the default ordering, yet an emergent reordering arises when a feasible gap opens while CAV4 yields for CAV1, CAV2 identifies the gap and commits to GO breaking strict FCFS order without any explicit reordering logic. Figure 3: Scenario 1 (all left turns): (a) ziz_i; (b) speeds Figure 4: Scenario 1(all left turns) snapshots: 4s and 6.5s This optimal order emerging is captured in Scenario 3 which is run under slightly perturbed initial position of CAV2, with the resulting opinion trajectories shown in Fig. 5 and Fig. 8 clearly showing a jump in decision making. Figure 5: Scenario 2(mixed manuevers): (a) ziz_i; (b) speeds. Figure 6: Scenario 2 (mixed) snapshots: 4s and 9.5s TABLE I: Scenario configurations and Last vehicle exit times (s), where CAV Entry: (in-lane, out-lane, did_i(m), viv_i(m/s)) Scenario CAV1 CAV2 CAV3 CAV4 FCFS (s) Proposed. (s) 1 (1, 6, 29, 7) (3, 8, 45, 8) (5, 2, 50, 8) (7, 4, 35, 7) 22.20 22.25 2 (1, 6, 29, 7) (3, 4, 45, 7) (5, 6, 50, 8) (7, 4, 35, 7) 17.09 17.20 3 (1, 6, 29, 7) (3, 4, 44, 7) (5, 6, 50, 8) (7, 4, 35, 7) 17.09 12.50 Figure 7: Scenario 3(mixed) snapshot at 6s Figure 8: Scenario 3(mixed maneuvers): (a) ziz_i; (b) speeds. IV-A Analysis Table I summarizes the last vehicle exit time across scenarios. In Scenario 1 and Scenario 2, the proposed method approximately equals the FCFS, consistent with the geometric structure of that scenario, where FCFS ordering is near-optimal and the opinion dynamics converge to the same sequence. While the table shows almost equal performance, it is worth noting that its achieved in a decentralized manner. Also, Scenario 3 shows how our method is adaptably efficient using scenario specific requirements and outperforms FCFS in every such interactive scenario. V Conclusion In this letter, we address the problem of decentralized intersection coordination through a signed-graph opinion dynamics framework, where a conflict topology network and a dynamic belief network jointly drive each vehicle to a GO or YIELD commitment without a central coordinator or solver. The formulation outperforms centralized approaches by allowing crossing order to emerge from geometric feasibility and arrival priority across crossing, merge, and mixed conflict scenarios through a single unified parameterization based solely on vehicle maneuver intentions. Future work will focus on extending the framework to incorporate uncertainty-aware belief updates, and validating the approach in mixed-traffic environments with higher density. References [1] A. Bizyaeva, A. Franci, and N. E. Leonard (2023) Nonlinear opinion dynamics with tunable sensitivity. IEEE Transactions on Automatic Control 68 (3), p. 1415–1430. External Links: Document Cited by: §I. [2] A. Bizyaeva, A. Franci, and N. E. Leonard (2025) Multitopic belief formation through bifurcations over signed social networks. IEEE Trans. Autom. Control 70 (8), p. 5082–5097. External Links: Document Cited by: §I, §I-A. [3] C. Cathcart, I. X. Belaustegui, A. Franci, and N. E. Leonard (2024) Spiking nonlinear opinion dynamics (s-nod) for agile decision-making. IEEE Control. Syst. Lett 8 (), p. 3267–3272. External Links: Document Cited by: §I-F. [4] C. Cathcart, M. Santos, S. Park, and N. E. Leonard (2023) Proactive opinion-driven robot navigation around human movers. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Vol. , p. 4052–4058. External Links: Document Cited by: §I. [5] X. Gong, B. Wang, and S. Liang (2024) Collision-free cooperative motion planning and decision-making for connected and automated vehicles at unsignalized intersections. IEEE Transactions on Systems, Man, and Cybernetics: Systems 54 (5), p. 2744–2756. Cited by: §I. [6] A. Katriniok, B. Rosarius, and P. Mähönen (2022) Fully distributed model predictive control of connected automated vehicles in intersections: theory and vehicle experiments. IEEE Trans. Intell. Transp. Syst 23 (10), p. 18288–18300. External Links: Document Cited by: §I. [7] M. Khayatian, M. Mehrabian, and A. Shrivastava (2018) RIM: robust intersection management for connected autonomous vehicles. In 2018 IEEE Real-Time Systems Symposium (RTSS), Vol. , p. 35–44. External Links: Document Cited by: §I. [8] Z. Lin, J. Lan, C. Anagnostopoulos, Z. Tian, and D. Flynn (2025) Safety-critical multi-agent mcts for mixed traffic coordination at unsignalized intersections. IEEE Trans. Intell. Transp. Syst. Cited by: §I. [9] X. Liu, P. Hang, Y. Wang, and J. Sun (2026) A cooperative decision-making method for cavs from the perspective of opinion dynamics. Transp. Res. Part C Emerg. Technol 182, p. 105412. Cited by: §I. [10] T. M. Paine and M. R. Benjamin (2024) A model for multi-agent autonomy that uses opinion dynamics and multi-objective behavior optimization. In 2024 IEEE International Conference on Robotics and Automation (ICRA), p. 8305–8311. Cited by: §I. [11] L. Riegger, M. Carlander, N. Lidander, N. Murgovski, and J. Sjöberg (2016) Centralized mpc for autonomous intersection crossing. In 2016 IEEE 19th International Conference on Intelligent Transportation Systems (ITSC), Vol. , p. 1372–1377. External Links: Document Cited by: §I. [12] T. Zaslavsky (2013) Matrices in the theory of signed simple graphs. arXiv preprint arXiv:1303.3083. Cited by: §I. [13] W. Zhong, K. Li, J. Shi, J. Yu, and Y. Luo (2024) Reservation-prioritization-based mixed-traffic cooperative control at unsignalized intersections. IEEE Transactions on Intelligent Vehicles 9 (5), p. 4917–4930. External Links: Document Cited by: §I.