{"id":"a1d38bc4-2dff-4330-b2db-ba8eab255e52","arxiv_id":"1908.05159","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A closed-form optimal acceleration profile for automated vehicles is derived that includes a speed-dependent rear-end safety constraint, extending the authors' prior unconstrained and state-constrained solutions.","lead":"This paper derives a closed-form optimal acceleration rule for automated vehicles that must keep a safe following distance while crossing an intersection. It extends the authors' earlier unconstrained solution by adding the rear-end safety constraint as a state constraint and demonstrating the resulting arc structure in simulation.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Completeness of the arc-switching enumeration in Sec. III is asserted, not proved; safety-active cases followed by v_max/u_max, simultaneous constraints, and repeated safety arcs are omitted, so the central closed-form claim is not established.","rationale":"I read the paper in good faith. It makes an incremental extension of the authors' prior decentralized optimal control framework by adding the rear-end safety constraint as an explicit state constraint and providing arc formulas. The three simulation cases illustrate the intended behavior, and the local optimality of each arc follows from standard calculus of variations. However, the strongest claim, that the solution is 'complete' and 'closed-form', depends on the assertion that the optimal trajectory always consists of the enumerated arcs in a single non-repeating sequence. The closing paragraph of Section III does not prove this; it merely says the process is repeated until no constraint is violated. This is a real gap: if a feasible optimal trajectory ever revisits a constraint or has two constraints active simultaneously, the formulas in Section III will not produce it. The numerical validation in Section IV tests only three particular scenarios and does not compare with an independent optimizer, so it cannot confirm completeness. I do not see a way to repair the claim without either a proof of the switching structure or a broad numerical validation. Thus the paper should remain conditional: the formulas are plausible and useful, but the central completeness claim should be either proved or explicitly weakened. This matches the reader's identified weakest assumption.","tokens_in":9393,"tokens_out":7632,"duration_ms":87045,"concrete_test":"Run a randomized numerical optimal-control comparison. Generate feasible instances with random initial gaps, speeds, exit times, and bounds in the simulation range, and with leading-vehicle velocity profiles v_k(t) that are non-monotone (accelerating then decelerating, and vice versa), including cases whose unconstrained solution would hit u_max or v_max after the safety constraint. For each instance, solve (2)-(6) with a direct collocation solver (e.g., CasADi/IPOPT or GPOPS-II), and independently run the Section III piecing algorithm using equations (14)-(26) and junction conditions. Report any instance where the piecing algorithm returns no feasible trajectory, returns a feasible trajectory with cost higher than the numerical optimum, or violates the necessary conditions at a supposed switching time.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that the optimal acceleration for (6) is a complete closed-form concatenation of free, safety-active, control-bound, and speed-bound arcs. The construction is a constructive enumeration: Section III gives formulas for each arc and says, in the closing paragraph, 'The process is repeated until the solution does not violate any other constraints.' This paragraph is the only evidence for exhaustiveness; no theorem states that the active set is visited in a single pass or that no two constraints can bind simultaneously. The enumerated cases are visibly incomplete: for the safety constraint, the only exit cases are to u_i,min (Case 1), to v_min (Case 2), or to a fresh unconstrained arc when v_k is increasing or constant. There is no case in which the safety-active arc exits to v_max or u_i,max, no case in which the safety constraint and a control or speed bound are active on a common interval, and no treatment of re-entry into the safety constraint after leaving it. Because the control on a safety-active arc is fixed by the condition \\dot{S}=0, namely u_i = ξ_i (v_k - v_i) / ρ_i, and the junction conditions (20)-(21) differ depending on which constraint is active after the exit, these missing cases change the algebraic equations that define the candidate. Without a proof that they cannot occur, the label 'complete' is unsupported. This is a correctness risk in the completeness claim, not a disagreement with the local optimality of the proposed arcs.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a lower-level optimal control problem for a connected and automated vehicle following a known preceding vehicle through a control zone. The dynamics are a double integrator augmented by a spacing state, with control limits, speed limits, and a rear-end safety constraint that is linear in speed. The authors claim a complete closed-form solution in which the optimal acceleration is a concatenation of unconstrained polynomial arcs and arcs where the safety constraint, control bound, or speed bound is active, with switching times determined by algebraic equations from boundary and interior conditions. Three simulation cases with different leading-vehicle acceleration profiles are presented to illustrate the construction.","tokens_in":9655,"tokens_out":3785,"duration_ms":34439,"significance":"If the closed-form solution is correct, it would be a useful building block for online decentralized CAV coordination: the solution is free of fitted parameters, the safe-distance constants and reaction constant are exogenous model inputs, and the unconstrained version has been experimentally validated in prior work by the authors. The paper makes a clear contribution by incorporating the rear-end safety constraint as a state constraint and providing explicit arc formulas. However, the central 'complete' claim rests on an unproved switching-structure enumeration, and Eq. (17) contains a sign error; these issues must be resolved before the result can be relied upon.","major_comments":[{"comment":"The paper's proof of completeness is a single sentence: 'The process is repeated until the solution does not violate any other constraints.' No argument is given that the active set sequence is single-pass, that no two constraints bind on a common interval, or that re-entry into an already-active constraint cannot occur. The enumerated cases are visibly incomplete: a safety-active arc never exits to v_max or u_i,max, there is no corner condition for simultaneous safety and control/speed constraints, and re-entry into the safety constraint is absent. Because the switching times are determined by the junction conditions (20)-(21), which differ by exit type, these missing configurations change the algebraic equations defining the candidate. The claim that the solution is 'complete' is therefore not established.","section":"Section III, closing paragraph"},{"comment":"The definition of N_i contains a sign inconsistency. With s_i = xi_i (p_k - p_i), the safety constraint is gamma_i + rho_i v_i - s_i = gamma_i + rho_i v_i - xi_i p_k + xi_i p_i, as in the first expression in (17). The immediately following expression, gamma_i + rho_i v_i - xi_i (p_k + p_i), has the wrong sign on p_i. This error propagates into the junction conditions (20)-(21) and should be corrected before the arc formulas are used.","section":"Equation (17)"},{"comment":"No sufficiency verification or independent numerical optimal control comparison is provided. The three simulations illustrate feasibility of the constructed arc sequence, but they do not demonstrate that the constructed trajectory minimizes (6) among all feasible controls, and they do not exercise the missing switching configurations identified above. A comparison against a numerical optimal control solver, such as direct collocation, would either corroborate the closed-form candidate or reveal the missing arc cases.","section":"Section IV and Section III"}],"minor_comments":[{"comment":"The phrase 'real-end collision avoidance' should read 'rear-end collision avoidance'.","section":"Section IV"},{"comment":"There is a typo: 'influence fundtions' should be 'influence functions', and the terms 'Lagrange multiplier' and 'Lagrange multiplier constant' are used inconsistently.","section":"Section III, around Eq. (20)"},{"comment":"The notation is inconsistent: 't_f^i' is sometimes written as 't_m^i', and the phrase 'assigned time for vehicle i to exits' should be corrected; these should be harmonized throughout.","section":"Section II"},{"comment":"The Hamiltonian introduces mu_c and mu_d for the speed bound constraints, but the text does not state which multiplier corresponds to which speed bound; a brief sentence defining these multipliers would improve clarity.","section":"Equation (7)"},{"comment":"The simulation description would be easier to reproduce if the parameter values (initial positions, speeds, xi_i, gamma_i, rho_i, v_min, v_max, u_i,min, u_i,max) were collected in a table, and if each subfigure in Figs. 2-4 were explicitly referenced in the text.","section":"Section IV"}],"recommendation":"major_revision","confidential_remarks":"The main technical issue is the missing completeness proof, not a hidden parameter-fitting problem: the derivation has no free fitted parameters, and the safety and control constants are exogenous. I would encourage the editor to require a rigorous proof of the switching-structure enumeration, a correction of the sign error in Eq. (17), and a numerical optimal control comparison before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new thing here is the extension of the authors' prior decentralized control framework to include a speed-dependent rear-end safety constraint as an explicit state constraint. For the cases they simulate, the arc construction is plausible and the formulas for the safety-active arc are genuinely new. No parameters are fitted; the safe-distance constants are exogenous inputs. Credit where due: the idea of augmenting the state with distance to the preceding vehicle and deriving the control on the safety-active arc from the condition dS/dt=0 is a sensible step that fills a real gap in the earlier solution.\n\nThe soft spots are several, and one is load-bearing. The paper claims a 'complete, closed-form analytical solution,' but the closing paragraphs of Section III merely assert that the process of piecing arcs is repeated until no constraint is violated. There is no proof that the enumerated active-set sequences are exhaustive. In particular, the safety-active arc is only ever exited to u_min, v_min, or a fresh unconstrained arc; exits to v_max or u_max are not treated, nor are intervals where the safety constraint and a control or speed bound bind simultaneously, nor re-entry into the safety constraint after leaving it. Without a proof that these cases cannot occur, the 'complete' label is unsupported. The stress-test note correctly identifies this.\n\nThere are also smaller issues that a careful revision should fix. Equation (17) has a sign error: the constraint should read ξ_i(p_k - p_i) for the distance, not ξ_i(p_k + p_i). The transversality condition (21) is stated with a partial derivative with respect to t1, which is at least nonstandard and likely another sign/notation slip. And while the authors compare against the unconstrained solution in simulation, they do not validate against a numerical optimal-control solver, so the reader cannot tell whether the pieced solution is actually optimal in cases not covered by the enumerated structure.\n\nThe derivation follows standard necessary conditions, and for the cases actually simulated the solution looks correct. But the central completeness claim is not established, and the sign error in (17) makes the junction analysis hard to trust without a corrected derivation.\n\nThis paper is for researchers working on real-time CAV trajectory planning who would value a closed-form inner-loop controller. It deserves a serious referee, but the referee should be asked to demand a rigorous treatment of the switching structure or an explicit statement that the solution is complete only under additional assumptions (e.g., monotonic leading-vehicle speed and no simultaneous constraints). I would not cite it in its current form, but I would look at a revised version.","headline":"This paper extends the authors' earlier Hamiltonian framework to include a rear-end safety constraint and offers explicit arc formulas, but the 'complete closed-form' claim rests on an unproven assertion about the switching structure.","tokens_in":10163,"tokens_out":1896,"would_cite":false,"duration_ms":20009,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["49K15","49J15","93C85"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper derives a complete closed-form optimal acceleration for connected vehicles that respects rear-end safety.","keywords":["connected and automated vehicles","optimal control","rear-end collision avoidance","closed-form solution","state constraints","Hamiltonian analysis","double integrator","L2 norm minimization"],"falsifier":"Construct a preceding-vehicle speed profile that alternates between strong deceleration and acceleration before the exit time, so that the following vehicle's safety constraint becomes active, inactive, and then active again. If the closed-form piecing procedure cannot produce a trajectory meeting the terminal position and all constraints, the claim of completeness fails.","tokens_in":9147,"feed_emoji":"🚗","tokens_out":2726,"duration_ms":30456,"temperature":0.7,"pith_summary":"The paper addresses the lower-level optimal control problem for a connected and automated vehicle crossing an intersection: given entry and exit times, initial state, and the trajectory of the vehicle ahead, find the acceleration of minimum L2 norm that obeys speed, acceleration, and rear-end safety constraints. Earlier work had solved the unconstrained problem and the problem with state and control constraints, but omitted the rear-end collision avoidance constraint. The authors claim to close that gap by augmenting the double-integrator vehicle model with a state for distance to the preceding vehicle, making safety a state constraint, and deriving a piecewise closed-form solution. If correct, each vehicle can compute its optimal acceleration online by solving algebraic equations for switching times, without numerical optimal control.","feed_headline":"Closed-form safe acceleration law derived for connected vehicles","feed_subtitle":"Each vehicle gets the exact minimum-energy acceleration arc by arc, without numerical optimal control.","key_machinery":"The central object is the augmented state vector $x_i = [p_i, v_i, s_i]^T$, where $s_i = \\xi_i (p_k - p_i)$ is the signed distance to the immediately preceding vehicle. This turns the rear-end safety constraint into a pure state constraint, $\\delta_i(t) - s_i(t) \\leq 0$ with $\\delta_i(t) = \\gamma_i + \\rho_i v_i(t)$. The argument then applies the standard Hamiltonian and Euler--Lagrange machinery, introducing Lagrange multipliers for the active arcs and corner conditions at junction points, and pieces together the admissible arcs to satisfy all interior and boundary conditions.","core_discovery":"For each vehicle with fixed entry and exit times, given initial states, and a known preceding-vehicle trajectory, the acceleration that minimizes the L2 norm subject to control bounds, speed bounds, and the rear-end safety constraint $s_i(t) \\geq \\gamma_i + \\rho_i v_i(t)$ is a concatenation of arcs: free driving, where acceleration is affine in time as in (14)--(16); a safety-active arc where $s_i(t) = \\gamma_i + \\rho_i v_i(t)$; a control-active arc where acceleration equals its minimum or maximum; and a speed-active arc where speed equals its lower bound. The arcs are pieced together at switching times determined by interior constraints and boundary conditions, and the switching structure is checked iteratively until no constraint is violated. The paper validates this construction in three simulation cases with different preceding-vehicle acceleration profiles.","pith_inferences":["A natural next test is whether the one-pass arc-enumeration remains complete when a constraint becomes active, then inactive, then active again within the same vehicle trajectory; the paper's closing argument does not explicitly rule out such repeated activation.","The same augmented-state technique could be extended to heterogeneous safety functions, e.g., nonlinear distance policies, though the resulting arcs would no longer be polynomial.","If the closed form extends to multiple vehicles in a platoon, the algebraic switching-time equations could be coupled through the preceding-vehicle trajectory, which may introduce new structural conditions not present in the single-follower case."],"forward_implications":["Each vehicle can compute its optimal acceleration profile in real time by solving a set of algebraic equations for the switching times, rather than running an iterative numerical optimizer.","Coordination frameworks that fix an intersection crossing schedule can now include rear-end safety in the lower-level energy-minimization step without changing the upper-level sequencing.","The safe-distance model with a speed-dependent headway term allows the method to adapt to varying safety policies across vehicles or scenarios.","The closed-form structure may serve as a baseline to benchmark numerical optimal control solvers in the same class of problems."],"supporting_citations":[{"why":"Provides the earlier analytical solution with state and control constraints but without the rear-end safety constraint, which this paper extends.","marker":"[19]"},{"why":"Establish the unconstrained double-integrator solution whose unconstrained arc is reused here.","marker":"[13]-[15]"},{"why":"Gives conditions under which the rear-end safety constraint never becomes active, framing the gap the paper fills.","marker":"[20]"},{"why":"Supplies the optimal-control theory for constrained arcs, corner conditions, and transversality used to piece the arcs.","marker":"[22]"},{"why":"Studies the same lower-level problem at adjacent intersections, providing a neighboring scenario where the new safety constraint could be applied.","marker":"[16]"}],"fun_headline_variants":["Exact closed-form acceleration for safe CAV coordination","Complete analytic solution includes rear-end safety constraint","Energy-optimal CAV arcs without numerical optimization","Closed-form switching law for safe connected vehicle motion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The solution assumes the optimal trajectory always follows the enumerated one-way sequence of arc types—free driving, then a safety-active interval, then a control-bound interval, then a speed-bound interval—so a trajectory that returns to an already-used constraint is not represented.","fun_headline_variants_meta":{"raw":{"variants":["Exact closed-form acceleration for safe CAV coordination","Complete analytic solution includes rear-end safety constraint","Energy-optimal CAV arcs without numerical optimization","Closed-form switching law for safe connected vehicle motion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000487,"raw_usage":{"total_tokens":2354,"prompt_tokens":851,"completion_tokens":1503,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":467,"completion_tokens_details":{"reasoning_tokens":1444}},"tokens_in":467,"tokens_out":1503,"duration_ms":14122,"temperature":1.0,"reasoning_tokens":1444,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:35:42.205398+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a preceding-vehicle speed profile that alternates between strong deceleration and acceleration before the exit time, so that the following vehicle's safety constraint becomes active, inactive, and then active again. If the closed-form piecing procedure cannot produce a trajectory meeting the terminal position and all constraints, the claim of completeness fails.","supporting_citations":[{"cited_title":"A decentralized energy-optimal control framework for connected automated vehicles at signal-free intersections,","cited_arxiv_id":null,"evidence_quote":"Provides the earlier analytical solution with state and control constraints but without the rear-end safety constraint, which this paper extends."},{"cited_title":"Opti- mal control for speed harmonization of automated vehicles,","cited_arxiv_id":null,"evidence_quote":"Gives conditions under which the rear-end safety constraint never becomes active, framing the gap the paper fills."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the optimal-control theory for constrained arcs, corner conditions, and transversality used to piece the arcs."},{"cited_title":"Optimal control and coordination of connected and automated vehicles at urban traffic intersections,","cited_arxiv_id":null,"evidence_quote":"Studies the same lower-level problem at adjacent intersections, providing a neighboring scenario where the new safety constraint could be applied."}],"review_version":1}