Skip to content

Test-driven Reinforcement Learning in Continuous Control

Conference: AAAI 2026 (Oral)
arXiv: 2511.07904
Code: https://github.com/KezhiAdore/TdRL
Area: Reinforcement Learning / Continuous Control / Task Representation
Keywords: Test-driven RL, Satisficing Theory, Multi-objective Optimization, Trajectory Return Function, Lexicographic Comparison

TL;DR

This paper proposes the Test-driven Reinforcement Learning (TdRL) framework, which replaces a single reward function with multiple test functions โ€” pass-fail tests defining optimality criteria and indicative tests guiding learning โ€” to represent task objectives. A return function is learned via lexicographic-heuristic trajectory comparison, matching or surpassing hand-crafted reward methods on the DeepMind Control Suite while naturally supporting multi-objective optimization.

Background & Motivation

In RL, the reward function bears a dual responsibility: (1) defining optimal behavior; and (2) guiding the learning process. This dual role makes reward design extremely difficult, frequently causing the following issues:

  • Reward Hacking: agents exploit loopholes in the reward function rather than genuinely completing the task
  • Design Bias: experts tend to evaluate individual state-action pairs in isolation, neglecting their effect on the entire trajectory
  • Multi-objective Dilemma: real-world tasks (e.g., autonomous driving must simultaneously account for safety, speed, comfort, and regulations) make it extremely difficult to determine weights among objectives
  • Limitations of PbRL: relies on human preference labels, which are subject to subjective bias
  • Limitations of IRL: requires large amounts of expert demonstrations and generalizes poorly
  • LLM-generated Rewards: still require manual domain knowledge and extensive iterative training feedback

Core Insight (Satisficing Theory): In multi-objective settings, humans do not pursue the optimum on a single metric but instead seek a "satisfactory solution" across objectives. For instance, drivers do not blindly minimize travel time but arrive on schedule subject to safety, comfort, and compliance constraints.

Method

Overall Architecture

A four-stage iterative pipeline: 1. Collect Trajectory: the policy interacts with the environment to collect trajectories 2. Return Learning: update the return function \(R_\xi^{ind}\) based on trajectory comparison outcomes 3. Reward Learning: decompose trajectory returns into a state-action reward function \(r_\phi(s,a)\) 4. Policy Optimization: optimize the policy using SAC/PPO based on the learned reward

Key Designs

  1. Test Function Taxonomy:

    • Pass-fail tests \(z^{pf}: \tau \to \{0, 1\}\): hard constraints defining optimal behavior (e.g., "did the agent reach the goal?" "is the torso upright?"), with binary outcomes
    • Indicative tests \(z^{ind}: \tau \to \mathbb{R}\): provide continuous metrics to guide learning (e.g., "locomotion speed," "energy consumption"); they do not define optimality but help discriminate between trajectories
    • The two test types have distinct roles: pass-fail tests define "what is correct," while indicative tests facilitate "how to learn efficiently"
    • No pre-specified weights are required for combination โ€” lexicographic ordering handles priority automatically
  2. Theoretical Guarantee (Theorem 1):

    • If the trajectory return function \(R(\tau)\) assigns higher returns to trajectories closer to the optimal trajectory set \(\tilde{\mathcal{T}}\) (distance monotonicity)
    • Then maximum-entropy policy optimization yields policies closer to the optimal policy set \(\tilde{\Pi}\)
    • Formally: \(d(\tau_1, \tilde{\mathcal{T}}) \leq d(\tau_2, \tilde{\mathcal{T}}) \Rightarrow R(\tau_1) \geq R(\tau_2) \Rightarrow d(\pi_2, \tilde{\Pi}) \leq d(\pi_1, \tilde{\Pi})\)
  3. Lexicographic Trajectory Comparison:

    • Rather than directly computing \(d(\tau, \tilde{\mathcal{T}})\) (since the optimal trajectory set is unknown), heuristic rules determine which of \(\tau_1\) and \(\tau_2\) is closer to \(\tilde{\mathcal{T}}\)
    • Priority rules: (1) both pass all pass-fail tests โ†’ tie; (2) the trajectory passing more pass-fail tests is preferred; (3) the trajectory passing harder tests (lower historical pass rate) is preferred; (4) rank by degree of optimization on indicative tests, prioritizing the least-optimized dimension
    • Outputs \(\mu \in \{0, 0.5, 1\}\) as preference labels for the Bradley-Terry model
  4. Return Function Learning:

    • A fully connected network \(R_\xi^{ind}\) maps \(n\) indicative test results to a scalar return
    • Loss: \(\mathcal{L}_R^{Dis}\) (distance-based cross-entropy) + \(\mathcal{L}_R^{Penalty}\) (numerical stability term)
    • Decomposition into state-action reward: \(\mathcal{L}_r = \sum[R(\tau) - \sum r_\phi(s,a)]^2\)
    • Two gradient balancing methods: GN (gradient norm normalization) and ES (early stopping, \(K^{ES}=10\))

Loss & Training

  • Return function: \(\mathcal{L}_R^{Dis}\) (BT-model cross-entropy) + \(\mathcal{L}_R^{Penalty}\) (MSE regularization)
  • Reward decomposition: \(\mathcal{L}_r = \text{MSE}(R(\tau),\, \sum r_\phi(s,a))\)
  • Policy optimization: standard SAC objective (with maximum entropy), or PPO

Key hyperparameters: 9,000 steps of unsupervised pre-training; trajectory buffer maximum capacity 100; segment size 50; reward/return network lr = \(3 \times 10^{-4}\); reward ensemble = 3

Key Experimental Results

Main Results: DM-Control Continuous Control Tasks

Task SAC + Oracle Reward TdRL-GN TdRL-ES PPO + Oracle PPO + TdRL
Walker-Stand ~980 โ‰ˆ980 โ‰ˆ980 ~970 ~960
Walker-Run ~650 โ‰ˆ670 โ‰ˆ650 ~550 ~480
Cheetah-Run ~830 โ‰ˆ830 โ‰ˆ850 ~750 ~720
Quadruped-Run ~780 โ‰ˆ800 โ‰ˆ770 ~650 ~600

Ablation Study

Variant Walker-Run Performance Notes
TdRL-ES (\(K^{ES}=10\)) ~650 Recommended default
Without Penalty term Unstable training Return values grow unbounded
Direct reward learning (no return decomposition) Unstable training Requires tanh clipping + repeated rescaling
\(K^{ES}\) too large Performance drops Insufficient Penalty constraint
\(K^{ES}\) too small Performance drops Over-constrained, return learning impeded

Multi-objective Analysis (Walker-Run)

Objective Oracle SAC TdRL Threshold
Torso upright \(\cos(\theta)\) โœ“ (satisfied) โœ“ (satisfied) [0.9, 1.0]
Torso height โœ— (not satisfied) โœ“ (satisfied) >1.2
X-axis velocity ~8 ~8 8

Key Findings

  • TdRL matches or surpasses oracle reward: achieves comparable performance without manual reward engineering
  • Natural multi-objective support: in Walker-Run, the weight design of the oracle reward leads to "upright but crouched running" (high upright score, low stand height); TdRL's pass-fail tests ensure all three objectives are satisfied
  • TdRL's policy stability is slightly lower than that of oracle reward โ€” trajectory-level evaluation inherently has greater variance than state-action evaluation, but this precisely mitigates reward hacking
  • Both GN and ES gradient balancing methods are effective; ES is simpler (recommended: \(K^{ES}=10\))
  • TdRL is also applicable to on-policy methods (PPO), although the theory is limited to the maximum-entropy framework

Highlights & Insights

  • A new task representation paradigm: test functions vs. reward functions, with functional separation (defining objectives vs. guiding learning)
  • The lexicographic comparison is inspired by human decision-making: hard constraints are checked first, then soft metrics โ€” consistent with the human heuristic of "ensure safety before seeking optimality"
  • Complete theoretical contribution: Theorem 1 provides convergence guarantees, rigorously derived from max-entropy RL
  • Key distinction from PbRL: TdRL does not require human preference labels but automatically generates trajectory comparisons via test functions
  • The conceptual contribution outweighs the engineering contribution โ€” the paper introduces a new way of thinking about RL task design

Limitations & Future Work

  • Theory is grounded solely in the maximum-entropy RL framework; on-policy methods such as PPO lack theoretical guarantees
  • Test functions still require manual design; though simpler than reward functions, the process is not fully automated
  • Lexicographic comparison is a heuristic; theoretical optimality is not guaranteed
  • Early-stage training is slower than with oracle reward (the return function must first be learned)
  • Validation is limited to simulated environments; no experiments on real robots or autonomous driving scenarios
Comparison Dimension TdRL PbRL IRL Reward Engineering
No preference labels required โœ“ โœ— โœ“ โœ“
No expert demonstrations required โœ“ โœ“ โœ— โœ“
Trajectory-level evaluation โœ“ โœ“ โ€” โœ—
Multi-objective support Native Requires design โ€” Requires weights
  • Future direction: using LLMs to automatically generate test functions (noted in the paper)
  • Broader potential of satisficing theory in RL

Rating

  • Novelty: โญโญโญโญโญ (entirely new paradigm; Oral-level conceptual contribution)
  • Technical Depth: โญโญโญโญ (theoretical theorem + heuristic algorithm + complete implementation)
  • Experimental Thoroughness: โญโญโญ (only 4 DM-Control tasks; real-world scenarios absent)
  • Practical Value: โญโญโญโญ (reduces task design complexity in multi-objective RL)
  • The DMC benchmark is relatively simple; scalability to more complex tasks remains to be verified
  • The lexicographic assumption presupposes a clear priority ordering among tests, which may be more complex in practice
  • The relationship with and advantages over RLHF require further clarification
  • vs. traditional reward design: simpler to design; objective definition and learning guidance are decoupled
  • vs. RLHF: replaces human comparative feedback with pre-defined test functions
  • vs. multi-objective RL: native framework support vs. requiring additional algorithmic adaptation

Inspiration & Connections

The test-driven task representation approach is applicable to safety constraint definition in autonomous driving. It also offers an alternative perspective on the reward hacking problem.

Rating โญโญโญโญ (4/5)

An Oral paper with novel concepts and solid theoretical contributions. However, the experimental scope is limited (DMC only), and test function design still involves subjective factors.