[論文レビュー] Hierarchical Task Model Predictive Control for Sequential Mobile Manipulation Tasks

Xintong Du, Siqi Zhou+1 more|arXiv (Cornell University)|Mar 10, 2026

Robot Manipulation and Learning被引用数 0

ひとこと要約

The paper introduces HTMPC, a hierarchical task model predictive control framework that coordinates base and arm tasks in sequence, improving efficiency and reactivity for sequential mobile manipulation tasks on a 9-DoF robot. It reformulates lexicographic optimization for online solvability and demonstrates superior performance over state-of-the-art methods.

ABSTRACT

Mobile manipulators are envisioned to serve more complex roles in people's everyday lives. With recent breakthroughs in large language models, task planners have become better at translating human verbal instructions into a sequence of tasks. However, there is still a need for a decision-making algorithm that can seamlessly interface with the high-level task planner to carry out the sequence of tasks efficiently. In this work, building on the idea of nonlinear lexicographic optimization, we propose a novel Hierarchical-Task Model Predictive Control framework that is able to complete sequential tasks with improved performance and reactivity by effectively leveraging the robot's redundancy. Compared to the state-of-the-art task-prioritized inverse kinematic control method, our approach has improved hierarchical trajectory tracking performance by 42% on average when facing task changes, robot singularity and reference variations. Compared to a typical single-task architecture, our proposed hierarchical task control architecture enables the robot to traverse a shorter path in task space and achieves an execution time 2.3 times faster when executing a sequence of delivery tasks. We demonstrated the results with real-world experiments on a 9 degrees of freedom mobile manipulator.

研究の動機と目的

• Motivate the need for reactive, efficient control for sequential mobile manipulation tasks guided by high-level task planners.
• Formulate sequential task execution as a Hierarchical Task Model Predictive Control problem.
• Leverage robot kinematic redundancy to coordinate base and end-effector tasks under lexicographic priorities.
• Demonstrate real-world performance gains on a 9-DoF mobile manipulator.
• Provide practical implementation guidance for online, iterative optimization in HTMPC.

提案手法

• Formulate the mobile manipulator dynamics as a linear-parameterized system with state x and input u; use accelerations as control inputs.
• Define a sequence of tracking tasks within an HTMPC framework that enforces a time-ordered hierarchy via lexicographic optimization over a prediction horizon.
• Solve a nested MPC where each single-task MPC (STMPC) optimizes a task with an added regularization and a relaxed lexicographic constraint to enable online solvability.
• Introduce two formulations of lexicographic constraints; opt for a decoupled constraint approximation to improve convergence (Eq. 12).
• Adopt a SQP-based solver with QP substeps and a line-search strategy; allow constrained relaxation to handle disturbances and model/measurement errors (slack/barrier techniques).
• Compare HTMPC against HTIDKC and a single-task HTMPC variant (HTMPC_WPT) using 9-DoF hardware, focusing on base/EE tracking and sequence completion performance.

実験結果

リサーチクエスチョン

• RQ1 階層的タスクモデル予測制御フレームワークは、従来の単一タスクや逆解運動学ベースの方法と比較して、連続的モバイルマニピュレーションタスクの効率と反応性を改善できるか？
• RQ2 MPC内での lexicographic 最適性の再定式化は、オンライン解法性とタスク変更・特異点・軌道変動下のタスク性能にどのように影響するか？
• RQ3 高自由度モバイルマニピュレータにおける階層制御アーキテクチャを活用した sequential delivery タスクの利点は何か？

主な発見

Eq	Low δ (mm) Avg	Low δ (mm) Std	High δ (mm) Avg	High δ (mm) Std
Baseline (11)	3.24	0.56	6.25	2.73
Proposed (12)	3.07	0.73	3.42	1.55

• HTMPCは、配送タスクの連続実行時に通常の単一タスクアーキテクチャよりも2.3倍速く実行できる。
• 最先端のタスク優先順位付き逆運動学制御法と比較して、HTMPCはタスク変更・特異点・軌道変動に直面した際の階層的軌道追従性能を平均約42%向上させる。
• HTMPCは、9-DoFのモバイルマニピュレータを用いた実世界の実験で、タスク空間の移動がより速く、効率的で、外乱からの回復性も高いことを示す。
• 2つのlexicographic formulationsを評価したところ、デカップルされた（Eq. 12）形式が収束と追従性能の点で直接的不等式アプローチ（Eq. 11）より優れていた。
• HTMPCは、連続タスクにおけるPart 2（valley-target）の追従性でHTIDKCおよびHTMPC_WPTより優れており、lexicographic適合性と反応性の改善を示す。